Driving Equitable Healthcare Innovations through an AI for Medicaid (AIM) Initiative
Artificial intelligence (AI) has transformative potential in the public health space – in an era when millions of Americans have limited access to high-quality healthcare services, AI-based tools and applications can enable remote diagnostics, drive efficiencies in implementation of public health interventions, and support clinical decision-making in low-resource settings. However, innovation driven primarily by the private sector today may be exacerbating existing disparities by training models on homogenous datasets and building tools that primarily benefit high socioeconomic status (SES) populations.
To address this gap, the Center for Medicare and Medicaid Innovation (CMMI) should create an AI for Medicaid (AIM) Initiative to distribute competitive grants to state Medicaid programs (in partnership with the private sector) for pilot AI solutions that lower costs and improve care delivery for rural and low-income populations covered by Medicaid.
Challenge & Opportunity
In 2022, the United States spent $4.5 trillion on healthcare, accounting for 17.3% of total GDP. Despite spending far more on healthcare per capita compared to other high-income countries, the United States has significantly worse outcomes, including lower life expectancy, higher death rates due to avoidable causes, and lesser access to healthcare services. Further, the 80 million low-income Americans reliant on state-administered Medicaid programs often have below-average health outcomes and the least access to healthcare services.
AI has the potential to transform the healthcare system – but innovation solely driven by the private sector results in the exacerbation of the previously described inequities. Algorithms in general are often trained on datasets that do not represent the underlying population – in many cases, these training biases result in tools and models that perform poorly for racial minorities, people living with comorbidities, and people of low SES. For example, until January 2023, the model used to prioritize patients for kidney transplants systematically ranked Black patients lower than White patients – the race component was identified and removed due to advocacy efforts within the medical community. AI models, while significantly more powerful than traditional predictive algorithms, are also more difficult to understand and engineer, resulting in the likelihood of further perpetuating such biases.
Additionally, startups innovating the digital health space today are not incentivized to develop solutions for marginalized populations. For example, in FY 2022, the top 10 startups focused on Medicaid received only $1.5B in private funding, while their Medicare Advantage (MA)-focused counterparts received over $20B. Medicaid’s lower margins are not attractive to investors, so digital health development targets populations that are already well-insured and have higher degrees of access to care.
The Federal Government is uniquely positioned to bridge the incentive gap between developers of AI-based tools in the private sector and American communities who would benefit most from said tools. Accordingly, the Center for Medicare and Medicaid Innovation (CMMI) should launch the AI for Medicaid (AIM) Initiative to incentivize and pilot novel AI healthcare tools and solutions targeting Medicaid recipients. Precedents in other countries demonstrate early success in state incentives unlocking health AI innovations – in 2023, the United Kingdom’s National Health Service (NHS) partnered with Deep Medical to pilot AI software that streamlines services by predicting and mitigating missed appointment risk. The successful pilot is now being adopted more broadly and is projected to save the NHS over $30M annually in the coming years.
The AIM Initiative, guided by the structure of the former Medicaid Innovation Accelerator Program (IAP), President Biden’s executive order on integrating equity into AI development, and HHS’ Equity Plan (2022), will encourage the private sector to partner with State Medicaid programs on solutions that benefit rural and low-income Americans covered by Medicaid and drive efficiencies in the overall healthcare system.
Plan of Action
CMMI will launch and operate the AIM Initiative within the Department of Health and Human Services (HHS). $20M of HHS’ annual budget request will be allocated towards the program. State Medicaid programs, in partnership with the private sector, will be invited to submit proposals for competitive grants. In addition to funding, CMMI will leverage the former structure of the Medicaid IAP program to provide state Medicaid agencies with technical assistance throughout their participation in the AIM Initiative. The programs ultimately selected for pilot funding will be monitored and evaluated for broader implementation in the future.
Sample Detailed Timeline
- 0-6 months:
- HHS Secretary to announce and launch the AI for Medicaid (AIM) Initiative within CMMI (e.g., delineating personnel responsibilities and engaging with stakeholders to shape the program)
- HHS to include AIM funding in annual budget request to Congress ($20M allocation)
- 6-12 months:
- CMMI to engage directly with state Medicaid agencies to support proposal development and facilitate connections with private sector partners
- CMMI to complete solicitation period and select ~7-10 proposals for pilot funding of ~$2-5M each by end of Year 1
- Year 2-7: Launch and roll out selected AI projects, led by state Medicaid agencies with continued technical assistance from CMMI
- Year 8: CMMI to produce an evaluative report and provide recommendations for broader adoption of AI tools and solutions within Medicaid-covered and other populations
Risks and Limitations
- Participation: Success of the initiative relies on state Medicaid programs and private sector partners’ participation. To mitigate this risk, CMMI will engage early with the National Association of Medicaid Directors (NAMD) to generate interest and provide technical assistance in proposal development. These conversations will also include input and support from the HHS Office of the Chief AI Officer (OCAIO) and its AI Council/Community of Practice. Further, startups in the healthcare AI space will be invited to engage with CMMI on identifying potential partnerships with state Medicaid agencies. A secondary goal of the initiative will be to ensure a number of private sector partners are involved in AIM.
- Oversight: AI is at the frontier of technological development today, and it is critical to ensure guardrails are in place to protect patients using AI technologies from potential adverse outcomes. To mitigate this risk, state Medicaid agencies will be required to submit detailed evaluation plans with their proposals. Additionally, informed consent and the ability to opt-out of data sharing when engaging with personally identifiable information (PII) and diagnostic or therapeutic technologies will be required. Technology partners (whether private, academic, or public sector) will further be required to demonstrate (1) adequate testing to identify and reduce bias in their AI tools to reasonable standards, (2) engagement with beneficiaries in the development process, and (3) leveraging testing environments that reflect the particular context of the Medicaid population. Finally, all proposals must adhere to guidelines published by AI guidelines adopted by HHS and the federal government more broadly, such as the CMS AI Playbook, the HHS Trustworthy AI Playbook, and any imminent regulations.
- Longevity: As a pilot grant program, the initiative does not promise long-term results for the broader population and will only facilitate short-term projects at the state level. Consequently, HHS leadership must remain committed to program evaluation and a long-term outlook on how AI can be integrated to support Americans more broadly. AI technologies or tools considered for acquisition by state Medicaid agencies or federal agencies after pilot implementation should ensure compliance with OMB guidelines.
Conclusion
The AI for Medicaid Initiative is an important step in ensuring the promise of artificial intelligence in healthcare extends to all Americans. The initiative will enable the piloting of a range of solutions at a relatively low cost, engage with stakeholders across the public and private sectors, and position the United States as a leader in healthcare AI technologies. Leveraging state incentives to address a critical market failure in the digital health space can additionally unlock significant efficiencies within the Medicaid program and the broader healthcare system. The rural and low-income Americans reliant on Medicaid have too often been an afterthought in access to healthcare services and technologies – the AIM Initiative provides an opportunity to address this health equity gap.
This action-ready policy memo is part of Day One 2025 — our effort to bring forward bold policy ideas, grounded in science and evidence, that can tackle the country’s biggest challenges and bring us closer to the prosperous, equitable and safe future that we all hope for whoever takes office in 2025 and beyond.
PLEASE NOTE (February 2025): Since publication several government websites have been taken offline. We apologize for any broken links to once accessible public data.
Strategies to Accelerate and Expand Access to the U.S. Innovation Economy
In 2020, we outlined a vision for how the incoming presidential administration could strengthen the nation’s innovation ecosystem, encouraging the development and commercialization of science and technology (S&T) based ventures. This vision entailed closing critical gaps from lab to market, with an emphasis on building a broadly inclusive pipeline of entrepreneurial talent while simultaneously providing key support in venture development.
During the intervening years, we have seen extraordinary progress, in good part due to ambitious legislation. Today, we propose innovative ways that the federal government can successfully build on this progress and make the most of new programs. With targeted policy interventions, we can efficiently and effectively support the U.S. innovation economy through the translation of breakthrough scientific research from the lab to the market. The action steps we propose are predicated on three core principles: inclusion, relevance, and sustainability. Accelerating our innovation economy and expanding access to it can make our nation more globally competitive, increase economic development, address climate change, and improve health outcomes. A strong innovation economy benefits everyone.
Challenge
Our Day One 2020 memo began by pitching the importance of innovation and entrepreneurship: “Advances in scientific and technological innovations—and, critically, the ability to efficiently transform breakthroughs into scalable businesses—have contributed enormously to American economic leadership over the past century.” Now, it is widely recognized that innovation and entrepreneurship are key to both global economic leadership and addressing the challenges of changing climate. The question is no longer whether we must innovate but rather how effectively we can stimulate and expand a national innovation economy.
Since 2020, the global and U.S. economies have gone through massive change and uncertainty. The Global Innovation Index (GII) 2023 described the challenges involved in its yearly analysis of monitoring global innovation trends amid uncertainty brought on by a sluggish economic recovery from the COVID-19 pandemic, elevated interest rates, and geopolitical tensions. Innovation indicators like scientific publications, research and development (R&D), venture capital (VC) investments, and the number of patents rose to historic levels, but the value of VC investment declined by close to 40%. As a counterweight to this extensive uncertainty, the GII 2023 described the future of S&T innovation and progress as “the promise of Digital Age and Deep Science innovation waves and technological progress.”
In the face of the pressures of global competitiveness, societal needs, and climate change, the clear way forward is to continue to innovate based on scientific and technical advancements. Meeting the challenges of our moment in history requires a comprehensive and multifaceted effort led by the federal government with many public and private partners.
Grow global competitiveness
Around the world, countries are realizing that investing in innovation is the most efficient way to transform their economies. In 2022, the U.S. had the largest R&D budget internationally, with spending growing by 5.6%, but China’s investment in R&D grew by 9.8%. For the U.S. to remain a global economic leader, we must continue to invest in innovation infrastructure, including the basic research and science, technology, engineering, and math (STEM) education that underpins our leadership, while we grow our investments in translational innovation. This includes reframing how existing resources are used as well as allocating new spending. It will require a systems change orientation and long-term commitments.
Increase economic development
Supporting and growing an innovation economy is one of our best tools for economic development. From place-based innovation programs to investment in emerging research institutions (ERIs) and Minority-Serving Institutions (MSIs) to training S&T innovators to become entrepreneurs in I-Corps™, these initiatives stimulate local economies, create high-quality jobs, and reinvigorate regions of the country left behind for too long.
Address climate change
In 2023, for the first time, global warming exceeded 1.5°C for an entire year. It is likely that all 12 months of 2024 will also exceed 1.5°C above pre-industrial temperatures. Nationally and internationally, we are experiencing the effects of climate change; climate mitigation, adaptation, and resilience solutions are urgently needed and will bring outsized economic and social impact.
Improve U.S. health outcomes
The COVID-19 pandemic was devastating, particularly impacting underserved and underrepresented populations, but it spurred unprecedented medical innovation and commercialization of new diagnostics, vaccines, and treatments. We must build on this momentum by applying what we’ve learned about rapid innovation to continue to improve U.S. health outcomes and to ensure that our nation’s health care needs across regions and demographics are addressed.
Make innovation more inclusive
Representational disparities persist across racial/ethnic and gender lines in both access to and participation in innovation and entrepreneurship. This is a massive loss for our innovation economy. The business case for broader inclusion and diversity is growing even stronger, with compelling data tracking the relationship between leadership diversity and company performance. Inclusive innovation is more effective innovation: a multitude of perspectives and lived experiences are required to fully understand complex problems and create truly useful solutions. To reap the full benefits of innovation and entrepreneurship, we must increase access and pathways for all.
Opportunity
With the new presidential administration in 2025, the federal government has a renewed opportunity to prioritize policies that will generate and activate a wave of powerful, inclusive innovation and entrepreneurship. Implementing such policies and funding the initiatives that result is crucial if we as a nation are to successfully address urgent problems such as the climate crisis and escalating health disparities.
Our proposed action steps are predicated on three core principles: inclusion, relevance, and sustainability.
Inclusion
One of this nation’s greatest and most unique strengths is our heterogeneity. We must leverage our diversity to meet the complexity of the substantial social and economic challenges that we face today. The multiplicity of our people, communities, identities, geographies, and lived experiences gives the U.S. an edge in the global innovation economy: When we bring all of these perspectives to the table, we better understand the challenges that we face, and we are better equipped to innovate to meet them. If we are to harness the fullness of our nation’s capacity for imagination, ingenuity, and creative problem-solving, entrepreneurship pathways must be inclusive, equitable, and accessible to all. Moreover, all innovators must learn to embrace complexity, think expansively and critically, and welcome perspectives beyond their own frame of reference. Collaboration and mutually beneficial partnerships are at the heart of inclusive innovation.
Relevance
Innovators and entrepreneurs have the greatest likelihood of success—and the greatest potential for impact—when their work is purpose-driven, nimble, responsive to consumer needs, and adaptable to different applications and settings. Research suggests that “breakthrough innovation” occurs when different actors bring complementary and independent skills to co-create interesting solutions to existing problems. Place-based innovation is one strategy to make certain that technology development is grounded in regional concerns and aspirations, leading to better outcomes for all concerned.
Sustainability
Multiple layers of sustainability should be integrated into the innovation and entrepreneurship landscape. First and most salient is supporting the development of innovative technologies that respond to the climate crisis and bolster national resilience. Second is encouraging innovators to incorporate sustainable materials and processes in all stages of research and development so that products benefit the planet and risks to the environment are mitigated through the manufacturing process, whether or not climate change is the focus of the technology. Third, it is vital to prioritize helping ventures develop sustainable business models that will result in long-term viability in the marketplace. Fourth, working with innovators to incorporate the potential impact of climate change into their business planning and projections ensures they are equipped to adapt to changing needs. All of these layers contribute to sustaining America’s social well-being and economic prosperity, ensuring that technological breakthroughs are accessible to all.
Proposed Action
Recommendation 1. Supply and prepare talent.
Continuing to grow the nation’s pipeline of S&T innovators and entrepreneurs is essential. Specifically, creating accessible entrepreneurial pathways in STEM will ensure equitable participation. Incentivizing individuals to become innovators-entrepreneurs, especially those from underrepresented groups, will strengthen national competitiveness by leveraging new, untapped potential across innovation ecosystems.
Expand the I-Corps model
By bringing together experienced industry mentors, commercial experts, research talent, and promising technologies, I-Corps teaches scientific innovators how to evaluate whether their innovation can be commercialized and how to take the first practical steps of bringing their product to market. Ten new I-Corps Hubs, launched in 2022, have expanded the network of engaged universities and collaborators, an important step toward growing an inclusive innovation ecosystem across the U.S.
Interest in I-Corps far outpaces current capacity, and increasing access will create more expansive pathways for underrepresented entrepreneurs. New federal initiatives to support place-based innovation and to grow investment at ERIs and MSIs will be more successful if they also include lab-to-market training programs such as I-Corps. Federal entities should institute policies and programs that increase awareness about and access to sequenced venture support opportunities for S&T innovators. These opportunities should include intentional “de-risking” strategies through training, advising, and mentoring.
Specifically, we recommend expanding I-Corps capacity so that all interested participants can be accommodated. We should also strive to increase access to I-Corps so that programs reach diverse students and researchers. This is essential given the U.S. culture of entrepreneurship that remains insufficiently inclusive of women, people of color, and those from low-income backgrounds, as well as international students and researchers, who often face barriers such as visa issues or a lack of institutional support needed to remain in the U.S. to develop their innovations. Finally, we should expand the scope of what I-Corps offers, so that programs provide follow-on support, funding, and access to mentor and investor networks even beyond the conclusion of initial entrepreneurial training.
I-Corps has already expanded beyond the National Science Foundation (NSF) to I-Corps at National Institutes of Health (NIH), to empower biomedical entrepreneurs, and Energy I-Corps, established by the Department of Energy (DOE) to accelerate the deployment of energy technologies. We see the opportunity to grow I-Corps further by building on this existing infrastructure and creating cohorts funded by additional science agencies so that more basic research is translated into commercially viable businesses.
Close opportunity gaps by supporting emerging research institutions (ERIs) and Minority-Serving Institutions (MSIs)
ERIs and MSIs provide pathways to S&T innovation and entrepreneurship, especially for individuals from underrepresented groups. In particular, a VentureWell-commissioned report identified that “MSIs are centers of research that address the unique challenges and opportunities faced by BIPOC communities. The research that takes place at MSIs offers solutions that benefit a broad and diverse audience; it contributes to a deeper understanding of societal issues and drives innovation that addresses these issues.”
The recent codification of ERIs in the 2022 CHIPS and Science Act pulls this category into focus. Defining this group, which comprises thousands of higher education institutions, was the first step in addressing the inequitable distribution of federal research funding. That imbalance has perpetuated regional disparities and impacted students from underrepresented groups, low-income students, and rural students in particular. Further investment in ERIs will result in more STEM-trained students, who can become innovators and entrepreneurs with training and engagement. Additional support that could be provided to ERIs includes increased research funding, access to capital/investment, capacity building (faculty development, student support services), industry partnerships, access to networks, data collection/benchmarking, and implementing effective translation policies, incentives, and curricula.
Supporting these institutions—many of which are located in underserved rural or urban communities that experience underinvestment—provides an anchor for sustained talent development and economic growth.
Recommendation 2. Support place-based innovation.
Place-based innovation not only spurs innovation but also builds resilience in vulnerable communities, enhancing both U.S. economic and national security. Communities that are underserved and underinvested in present vulnerabilities that hostile actors outside of the U.S. can exploit. Place-based innovation builds resilience: innovation creates high-quality jobs and brings energy and hope to communities that have been left behind, leveraging the unique strengths, ecosystems, assets, and needs of specific regions to drive economic growth and address local challenges.
Evaluate and learn from transformative new investments
There have been historic levels of government investment in place-based innovation, funding the NSF’s Regional Innovation Engines awards and two U.S. Department of Commerce Economic Development Administration (EDA) programs: the Build Back Better Regional Challenge and Regional Technology and Innovation Hubs awards. The next steps are to refine, improve, and evaluate these initiatives as we move forward.
Unify the evaluation framework, paired with local solutions
Currently, evaluating the effectiveness and outcomes of place-based initiatives is challenging, as benchmarks and metrics can vary by region. We propose a unified framework paired with solutions locally identified by and tailored to the specific needs of the regional innovation ecosystem. A functioning ecosystem cannot be simply overlaid upon a community but must be built by and for that community. The success of these initiatives requires active evaluation and incorporation of these learnings into effective solutions, as well as deep strategic collaboration at the local level, with support and time built into processes.
Recommendation 3. Increase access to financing and capital.
Funding is the lifeblood of innovation. S&T innovation requires more investment and more time to bring to market than other types of ventures, and early-stage investments in S&T startups are often perceived as risky by those who seek a financial return. Bringing large quantities of early-stage S&T innovations to the point in the commercialization process where substantial private capital takes an interest requires nondilutive and patient government support. The return on investment that the federal government seeks is measured in companies successfully launched, jobs created, and useful technologies brought to market.
Disparities in access to capital by companies owned by women and underrepresented minority founders are well documented. The federal government has an interest in funding innovators and entrepreneurs from many backgrounds: they bring deep and varied knowledge and a multitude of perspectives to their innovations and to their ventures. This results in improved solutions and better products at a cheaper price for consumers. Increasing access to financing and capital is essential to our national economic well-being and to our efforts to build climate resilience.
Expand SBIR/STTR access and commercial impact
The SBIR and STTR programs spur innovation, bolster U.S. economic competitiveness, and strengthen the small business sector, but barriers persist. In a recent third-party assessment of the SBIR/STTR program at NIH, the second largest administrator of SBIR/STTR funds, the committee found outreach from the SBIR/STTR programs to underserved groups is not coordinated, and there has been little improvement in the share of applications from or awards to these groups in the past 20 years. Further, NIH follows the same processes used for awarding R01 research grants, using the same review criteria and typically the same reviewers, omitting important commercialization considerations.
To expand access and increase the commercialization potential of the SBIR/STTR program, funding agencies should foster partnerships with a broader group of organizations, conduct targeted outreach to potential applicants, offer additional application assistance to potential applicants, work with partners to develop mentorship and entrepreneur training programs, and increase the percentage of private-sector reviewers with entrepreneurial experience. Successful example programs of SBIR/STTR support programs include the NSF Beat-The-Odds Boot Camp, Michigan’s Emerging Technologies Fund, and the SBIR/STTR Innovation Summit.
Provide entrepreneurship education and training
Initiatives like NSF Engines, Tech Hubs, Build-Back-Better Regional Challenge, the Minority Business Development Agency (MBDA) Capital Challenge, and the Small Business Administration (SBA) Growth Accelerator Fund expansion will all achieve more substantial results with supplemental training for participants in how to develop and launch a technology-based business. As an example of the potential impact, more than 2,500 teams have participated in I-Corps since the program’s inception in 2012. More than half of these teams, nearly 1,400, have launched startups that have cumulatively raised $3.16 billion in subsequent funding, creating over 11,000 jobs. Now is an opportune moment to widely apply similarly effective approaches.
Launch a local investment education initiative
Angel investors are typically providing the first private funding available to S&T innovators and entrepreneurs. These very early-stage funders give innovators access to needed capital, networks, and advice to get their ventures off the ground. We recommend that the federal government expand the definition of an accredited investor and incentivize regionally focused initiatives to educate policymakers and other regional stakeholders about best practices to foster more diverse and inclusive angel investment networks. With the right approach and support, there is the potential to engage thousands more high-net-worth individuals in early-stage investing, contributing their expertise and networks as well as their wealth.
Encourage investment in climate solutions
Extreme climate-change-attributed weather events such as floods, hurricanes, drought, wildfire, and heat waves cost the global economy an average of $143 billion annually. S&T innovations have the potential to help address the impacts of climate change at every level:
- Mitigation. Promising new ideas and technologies can slow or even prevent further climate change by reducing or removing greenhouse gasses.
- Adaptation. We can adapt processes and systems to better respond to adverse events, reducing the impacts of climate change.
- Resilience. By anticipating, preparing for, and responding to hazardous events, trends, or disturbances caused by climate change, we can continue to thrive on our changing planet.
Given the global scope of the problem and the shared resources of affected communities, the federal government can be a leader in prioritizing, collaborating, and investing in solutions to direct and encourage S&T innovation for climate solutions. There is no question whether climate adaptation technologies will be needed, but we must ensure that these solutions are technologies that create economic opportunity in the U.S. We encourage the expansion and regular appropriations of funding for successful climate programs across federal agencies, including the DoE Office of Technology Transitions’ Energy Program for Innovation Clusters, the National Oceanic and Atmospheric Administration’s (NOAA) Ocean-Based Climate Resilience Accelerators program, and the U.S. Department of Agriculture’s Climate Hubs.
Recommendation 4. Shift to a systems change orientation.
To truly stimulate a national innovation economy, we need long-term commitments in policy, practice, and regulations. Leadership and coordination from the executive branch of the federal government are essential to continue the positive actions already begun by the Biden-Harris Administration.
These initiatives include:
- Scientific integrity and evidence-based policy-making memo
- Catalyzing Clean Energy Industry Executive Order
- Implementation of the Infrastructure Investment and Jobs Act
- Advancing Biotechnology and Biomanufacturing Innovation for a Sustainable, Safe, and Secure American Bioeconomy
- Implementation of the CHIPS Act of 2022
- Advancing Women’s Health Research and Innovation
Policy
Signature initiatives like the CHIPS and Science Act, Infrastructure Investment and Jobs Act, and the National Quantum Initiative Act are already threatened by looming appropriations shortfalls. We need to fully fund existing legislation, with a focus on innovative and translational R&D. According to a report by PricewaterhouseCoopers, if the U.S. increased federal R&D spending to 1% of GDP by 2030, the nation could support 3.4 million jobs and add $301 billion in labor income, $478 billion in economic value, and $81 billion in tax revenue. Beyond funding, we propose supporting innovative policies to bolster U.S. innovation capacity at the local and national levels. This includes providing R&D tax credits to spur research collaboration between industry and universities and labs, providing federal matching funds for state and regional technology transfer and commercialization efforts, and revising the tax code to support innovation by research-intensive, pre-revenue companies.
Practice
The University and Small Business Patent Procedures Act of 1980, commonly known as the Bayh-Dole Act, allows recipients of federal research funding to retain rights to inventions conceived or developed with that funding. The academic tech transfer system created by the Bayh-Dole Act (codified as amended at 35 U.S.C. §§ 200-212) generated nearly $1.3 trillion in economic output, supported over 4.2 million jobs, and launched over 11,000 startups. We should preserve the Bayh-Dole Act as a means to promote commercialization and prohibit the consideration of specific factors, such as price, in march-in determinations.
In addition to the continual practice and implementation of successful laws such as Bayh-Dole, we must repurpose resources to support innovation and the high-value jobs that result from S&T innovation. We believe the new administration should allocate a share of federal funding to promote technology transfer and commercialization and better incentivize commercialization activities at federal labs and research institutes. This could include new programs such as mentoring programs for researcher entrepreneurs and student entrepreneurship training programs. Incentives include evaluating the economic impact of lab-developed technology by measuring commercialization outcomes in the annual Performance Evaluation and Management Plans of federal labs, establishing stronger university entrepreneurship reporting requirements to track and reward universities that create new businesses and startups, and incentivizing universities to focus more on commercialization activities as part of promotion and tenure of faculty,
Regulations
A common cause of lab-to-market failure is the inability to secure regulatory approval, particularly for novel technologies in nascent industries. Regulation can limit potentially innovative paths, increase innovation costs, and create a compliance burden on businesses that stifle innovation. Regulation can also spur innovation by enabling the management of risk. In 1976 the Cambridge (Massachusetts) City Council became the first jurisdiction to regulate recombinant DNA, issuing the first genetic engineering license and creating the first biotech company. Now Boston/Cambridge is the world’s largest biotech hub: home to over 1,000 biotech companies, 21% of all VC biotech investments, and 15% of the U.S. drug development pipeline.
To advance innovation, we propose two specific regulatory actions:
- Climate. We recommend the Environmental Protection Agency (EPA) adopt market-based strategies to help fight climate change by monitoring and regulating CO2 emissions, putting an explicit price on carbon emissions, and incentivizing businesses to find cost-effective and innovative ways to reduce those emissions.
- Health. We recommend strengthening regulatory collaboration between the Food and Drug Administration (FDA) and the Centers for Medicare & Medicaid Services (CMS) to establish a more efficient and timely reimbursement process for novel FDA-authorized medical devices and diagnostics. This includes refining the Medicare Coverage of Innovative Technologies rule and fully implementing the new Transitional Coverage for Emerging Technologies pathway to expedite the review, coverage determination, and reimbursement of novel medical technologies.
Conclusion
To maintain its global leadership role, the United States must invest in the individuals, institutions, and ecosystems critical to a thriving, inclusive innovation economy. This includes mobilizing access, inclusion, and talent through novel entrepreneurship training programs; investing, incentivizing, and building the capacity of our research institutions; and enabling innovation pathways by increasing access to capital, networks, and resources.
Fortunately, there are several important pieces of legislation recommitting the U.S. leadership to bold S&T goals, although much of the necessary resources are yet to be committed to those efforts. As a society, we benefit when federally supported innovation efforts tackle big problems that are beyond the scope of single ventures; notably, the many challenges arising from climate change. A stronger, more inclusive innovation economy benefits the users of S&T-based innovations, individual innovators, and the nation as a whole.
When we intentionally create pathways to innovation and entrepreneurship for underrepresented individuals, we build on our strengths. In the United States, our strength has always been our people, who bring problem-solving abilities from a multitude of perspectives and settings. We must unleash their entrepreneurial power and become, even more, a country of innovators..
Earlier memo contributors Heath Naquin and Shaheen Mamawala (2020) were not involved with this 2024 memo.
This action-ready policy memo is part of Day One 2025 — our effort to bring forward bold policy ideas, grounded in science and evidence, that can tackle the country’s biggest challenges and bring us closer to the prosperous, equitable and safe future that we all hope for whoever takes office in 2025 and beyond.
PLEASE NOTE (February 2025): Since publication several government websites have been taken offline. We apologize for any broken links to once accessible public data.
Collaborative Intelligence: Harnessing Crowd Forecasting for National Security
“The decisions that humans make can be extraordinarily costly. The wars in Iraq and Afghanistan were multi-trillion dollar decisions. If you can improve the accuracy of forecasting individual strategies by just a percentage point, that would be worth tens of billions of dollars.” – Jason Matheny, CEO, RAND Corporation
Predicting the future—a notoriously hard problem—is a core function of the Office of the Director of National Intelligence (ODNI). Crowd forecasting methods offer a systematic approach to quantifying the U.S. intelligence community’s uncertainty about the future and predicting the impact of interventions, allowing decision-makers to strategize effectively and allocate resources by outlining risks and tradeoffs in a legible format. We propose that ODNI leverage its earlier investments in crowd-forecasting research to enhance intelligence analysis and interagency coordination. Specifically, ODNI should develop a next-generation crowd-forecasting program that balances academic rigor with policy relevance. To do this, we propose partnering a Federally Funded Research and Development Center (FFRDC) with crowd forecasting experience with executive branch agencies to generate high-value forecasting questions and integrate targeted forecasts into existing briefing and decision-making processes. Crucially, end users (e.g. from the NSC, DoD, etc.) should be embedded in the question-generation process in order to ensure that the forecasts are policy-relevant. This approach has the potential to significantly enhance the quality and impact of intelligence analysis, leading to more robust and informed national security decisions.
Challenge & Opportunity
ODNI is responsible for the daunting task of delivering insightful, actionable intelligence in a world of rapidly evolving threats and unprecedented complexity. Traditional analytical methods, while valuable, struggle to keep pace with the speed and intricacy of global events where dynamic reports are necessary. Crowd forecasting provides infrastructure for building shared understanding across the Intelligence Community (IC) with a very low barrier to entry. Through the process, each agency can share their assessments of likely outcomes and planned actions based on their intelligence, to be aggregated alongside other agencies. These techniques can serve as powerful tools for interagency coordination within the IC, quickly surfacing areas of consensus and disagreement. By building upon the foundation of existing Intelligence Advanced Research Projects Activity (IARPA) crowd forecasting research — including IARPA’s Aggregative Contingent Estimation (ACE) tournament and Hybrid Forecasting Competition (HFC) — ODNI has within its reach significant low-hanging fruit for improving the quality of its intelligence analysis and the use of this analysis to inform decision-making.
Despite the IC’s significant investment in research demonstrating the potential of crowd forecasting, integrating these approaches into decision-making processes has proven difficult. The first-generation forecasting competitions showed significant returns from basic cognitive debiasing training, above and beyond the benefits of crowd forecast aggregation. Yet, attempts to incorporate forecasting training and probabilistic estimates into intelligence analysis have fallen flat due in large part to internal politics. Accordingly, the incentives within and among agencies must be considered in order for any forecasting program to deliver value. Importantly, any new crowd forecasting initiative should be explicitly rolled out as a complement, not a substitute, to traditional intelligence analysis.
Plan of Action
The incoming administration should direct the Office of the Director of National Intelligence (ODNI) to resume its study and implementation of crowd forecasting methods for intelligence analysis. The following recommendations illustrate how this can be done effectively.
Recommendation 1. Develop a Next-Generation Crowd Forecasting Program
Direct a Federally Funded Research and Development Center (FFRDC) experienced with crowd forecasting methods, such as MITRE’s National Security Engineering Center (NSEC) or the RAND Forecasting Initiative (RFI), to develop a next-generation pilot program.
Prior IARPA studies of crowd-sourced intelligence were focused on the question: How accurate is the wisdom of the crowds on geopolitical questions? To answer this, the IARPA tournaments posed many forecasting questions, rapid-fire, over a relatively short period of time, and these questions were optimized for easy generation and resolution (i.e. straightforward data-driven questions) — at the expense of policy relevance. A next-generation forecasting program should build upon recent research on eliciting from experts the crucial questions that illuminate key uncertainties, point to important areas of disagreement, and estimate the impact of interventions under consideration.
This program should:
- Incorporate lessons learned from previous IARPA forecasting tournaments, including difficulties with getting buy-in from leadership to incentivize the participation of busy analysts and decision-makers at ODNI.
- Develop a framework for generating questions that balance rigor, resolvability, and policy relevance.
- Implement advanced aggregation and scoring methods, leveraging recent academic research and machine learning methods.
Recommendation 2. Embed the Decision-Maker in the Question Generation Process
Direct the FFRDC to work directly with one or more executive branch partners to embed end users in the process of eliciting policy-relevant forecasting questions. Potential executive branch partners could include the National Security Council, Department of Defense, Department of State, and Department of Homeland Security, among others.
A formal process for question generation and refinement should be established, which could include:
- A structured methodology for transforming policy questions of interest into specific, quantifiable forecasting questions.
- A review process to ensure that questions meet criteria for both forecasting suitability and policy relevance.
- Mechanisms for rapid question development in response to emerging crises or sudden shifts.
- Feedback mechanisms to refine and improve question quality over time, with a focus on policy relevance and decision-maker user experience.
Recommendation 3. Integrate Forecasts into Decision-Making Processes
Ensure that resulting forecasts are actively reviewed by decision-makers and integrated into existing intelligence and policy-making processes.
This could involve:
- Incorporating forecast results into regular intelligence briefings, as a quantitative supplement to traditional qualitative assessments.
- Developing visualizations/dashboards (Figure 1) to enable decision-makers to explore the reasoning, drivers of disagreement, unresolved uncertainties and changes in forecasts over time.
- Organizing training sessions for senior leadership on how to interpret and use probabilistic forecasts in decision-making.
- Establishing a simple, formal process by which policymakers can request forecasts on questions relevant to their work.
- Creating a review process to assess how forecasts influenced decisions and their outcomes.
- Using forecast as a tool for interagency coordination, to surface ideas and concerns that people may be hesitant to bring up in front of their superiors.

Figure 1. Example of prototype forecasting dashboards for end-users, highlighting key factors and showing trends in the aggregate forecast over time. (Source: Metaculus)
Conclusion
ODNI’s mission to “deliver the most insightful intelligence possible” demands continuous innovation. The next-generation forecasting program outlined in this document is the natural next step in advancing the science of forecasting to serve the public interest. Crowd forecasting has proven itself as a generator of reliable predictions, more accurate than any individual forecaster. In an increasingly complex information environment, our intelligence community needs to use every tool at its disposal to identify and address its most pressing questions about the future. By establishing a transparent and rigorous crowd-forecasting process, ODNI can harness the collective wisdom of diverse experts and analysts and foster better interagency collaboration, strengthening our nation’s ability to anticipate and respond to emerging global challenges.
This action-ready policy memo is part of Day One 2025 — our effort to bring forward bold policy ideas, grounded in science and evidence, that can tackle the country’s biggest challenges and bring us closer to the prosperous, equitable and safe future that we all hope for whoever takes office in 2025 and beyond.
PLEASE NOTE (February 2025): Since publication several government websites have been taken offline. We apologize for any broken links to once accessible public data.
The Energy Transition Workforce Initiative
The energy transition underway in the United States continues to present a unique set of opportunities to put Americans back to work through the deployment of new technologies, infrastructure, energy efficiency, and expansion of the electricity system to meet our carbon goals. Unlike many previous industrial transitions, the U.S. can directly influence the pace of change, promote greater social equity, and create new jobs to replace those that are phasing out.
Since 2021, significant policies have been enacted to support this transition, including the Infrastructure Investment and Jobs Act (IIJA), CHIPS and Science Act, and the Inflation Reduction Act. The most recent Congressional Budget Office estimates of the energy-related spending of these three pieces of legislation was at least $956 billion over a 10-year period.
Despite these historic investments, additional work remains to be done. To supplement the accomplishments of the last four years, the next administration should:
- Establish the Energy Workforce and Economic Development Extension Program inside the Department of Energy (DOE).
- Restore the interagency Energy and Advanced Manufacturing Workforce Initiative.
- Initiate the Energy Transition Community Benefits Training Program.
- Establish a national public-private commission on steel decarbonization.
- Restore the DOE Labor Working Group under the direction of a senior advisor to the Secretary of Energy.
Challenge and Opportunity
In 2023, the energy sector added over 250,000 jobs, with clean energy accounting for 56% of jobs. Energy efficiency jobs, such as the manufacture and installation of heat pumps, added 74,700 jobs, the most of any technology area. While energy jobs are found in every state in America, fossil fuel production jobs and the infrastructure associated with them are highly concentrated. In 2020, 73% of the roughly one million oil, coal, and natural gas production jobs were in just 10 states. By 2023, 70,000 of those jobs were lost in the same 10 states, leaving the communities that host them at risk of economic decline. The Interagency Working Group on Coal and Power Plant Communities was established by Executive Order in 2021 to address this issue and provide new incentives for clean energy production such as the Sparkz and Form Energy battery plants in West Virginia. To date, over $538 billion of competitive and formula funding has been provided to “revitalize America’s energy communities.”
Plan of Action
On day one, the next administration should announce the expansion of the DOE Office of Energy Jobs to lead the following efforts.
Recommendation 1. Establish the Energy Workforce and Economic Development Extension Program (EWEDEP) inside the DOE.
Modeled after the Agricultural Extension Program, and in partnership with the National Laboratories, the EWEDEP should provide technical advice to the state decarbonization plans funded by the Environmental Protection Agency, as well as to municipalities, regional entities, tribal governments, and private-sector businesses. Led by the Office of Energy Jobs, this program should also assist regional, state, local, and tribal governments in developing and implementing technical decarbonization strategies that simultaneously create good local jobs. State and regional support staff for the Office of Energy Jobs should be located in each of the national laboratories.
Recommendation 2. Restore the interagency Energy and Advanced Manufacturing Workforce Initiative (EAMWI).
During the Obama Administration, EAMWI, run by the Department of Energy, coordinated activities between the Departments of Energy, Labor, Education, Commerce, and Defense and the National Science Foundation to harmonize planning, training, and curriculum development for the new energy workforce. In addition to resuming those coordinative activities, the next administration should mandate that the EAMWI produce quarterly assessments of the needs and opportunities in workforce training in response to the requirements of the energy transition. Based on updated USEER data from 2024 and ongoing job occupational needs’ assessments, EAMWI should provide annual reports on state energy workforce needs to the appropriate federal and state agencies in charge of energy, education, and economic development strategies.
Recommendation 3. Initiate the Energy Transition Community Benefits Training Program.
Community Benefit Plans (CBPs) and Community Benefit Agreements (CBAs) have emerged as the primary tools for monitoring job quality metrics in the energy transition, particularly those that are supported by federal government grants and loans. This program should provide expert training in the design and performance of CBPs and CBAs for company executives, community organizations and advocates, labor unions, and local government employees. This program should be informed by an advisory board of experts from business schools, trade associations, labor unions, and community stakeholders.
Recommendation 4. Establish a national public-private commission on steel decarbonization.
Decarbonizing the steel industry will be one of the most difficult and expensive challenges posed on the energy transition. Appointing a national commission of industry stakeholders, including business, labor, communities, and federal agencies, will be critical for developing a model for managing hard-to-decarbonize, industrial sectors of the economy in ways that create quality jobs, protect communities, and build broad consensus among the American people. DOE should also establish an Office of Steel Decarbonization to implement the commission’s recommendations.
Recommendation 5. Restore the DOE Labor Working Group under the direction of a senior advisor to the Secretary of Energy.
The DOE Labor Working Group provided monthly guidance on how to implement high wage strategies in the energy sector while preserving jobs and reducing greenhouse gas emissions. Member organizations included energy sector unions involved in the mining, extraction, manufacturing, construction, utility, and transportation industry sectors.
After initiating these actions on day one, the next administration should prioritize legislation establishing an Energy Transition Adjustment Assistance Program (ETAAP). In some cases, the loss of fossil fuel jobs in concentrated parts of the country will require retraining of current employees to prepare them for new careers with new employers. The U.S. will need a program to provide income support greater than extended unemployment to recipients undergoing retraining. Such a program should learn from the shortcomings of the Trade Adjustment Assistance (TAA) program by providing more supportive services. Based on two-year training costs and average participation rates of TAA-certified beneficiaries, a minimum of $20 billion for worker retraining should be allocated as part of this effort.
In addition, the Interagency Working Group on Coal and Power Plant Communities should be consulted to design standards for broad eligibility to participate in the ETAAP, including energy-intensive manufacturing businesses impacted by the energy transition. Finally, as existing energy companies transition to producing cleaner forms of energy, the program should consider subsidizing the retraining of existing energy-sector employees to provide new skills for the transition.
Conclusion
Unlike many previous industrial transitions, which were driven by new technologies and market forces, decarbonization is driven largely by social policy interventions. Thus, well-planned responses, based on timely clean-energy economic development investments, can provide good jobs and economic opportunity for displaced workers and affected communities. The clean energy tax credits included in the IRA should be maintained and extended. Labor standards and domestic content rules should be attached to both grants and formula spending. Finally, the lending authorities for the DOE Loan Program Office should be expanded to include energy infrastructure, energy-intensive manufacturing, and energy efficiency projects. With such an approach, the U.S. and its workers can benefit from the global push to decarbonize.
This idea was originally published on February 1, 2021. We’ve republished this updated version on November 27, 2024.
This action-ready policy memo is part of Day One 2025 — our effort to bring forward bold policy ideas, grounded in science and evidence, that can tackle the country’s biggest challenges and bring us closer to the prosperous, equitable and safe future that we all hope for whoever takes office in 2025 and beyond.
PLEASE NOTE (February 2025): Since publication several government websites have been taken offline. We apologize for any broken links to once accessible public data.
The main challenge is providing a timely economic development response to impacted communities before the most serious job losses have occurred. Our goal is to create a Federal Emergency Management Agency (FEMA)-like response in advance of the economic storm devastating some communities because of the loss of fossil fuel jobs. However, unlike FEMA, most federal economic development programs are not designed to respond to emergency job loss, and they require annual appropriations and lengthy preparations.
The overall success of the Energy Transition Workforce Initiative will be measured by the number and quality of jobs created in the communities expected to be hardest hit by the energy transition, the timeliness of the intervention, and the stability of the communities. Utilization rates of EWEDEP technical support for regions, state, local and tribal governments to develop implementation plans will also be a primary measure.
Promoting Fusion Energy Leadership with U.S. Tritium Production Capacity
As a fusion energy future becomes increasingly tangible, the United States should proactively prepare for it if/when it arrives. A single, commercial-scale fusion reactor will require more tritium fuel than is currently available from global civilian-use inventories. For fusion to be viable, greater-than-replacement tritium breeding technologies will be essential. Before the cycle of net tritium gain can begin, however, the world needs sufficient tritium to complete R&D and successfully commission first-of-a-kind (FOAK) fusion reactors. The United States has the only proven and scalable tritium production supply chain, but it is largely reserved for nuclear weapons. Excess tritium production capacity should be leveraged to ensure the success of and U.S. leadership in fusion energy.
The Trump administration should reinforce U.S. investments and leadership in commercial fusion with game changing innovation in the provision of tritium fuel. The Congressional Fusion Energy Caucus has growing support in the House with 92 members and an emerging Senate counterpart chaired by Sen. Martin Heinrich. Energy security and independence are important areas of bipartisan cooperation, but strong leadership from the White House will be needed to set a bold, America-first agenda.
Challenge and Opportunity
Fusion energy R&D currently relies on limited reserves of tritium from non-scalable production streams. These reserves reduce by ~5% each year due to radioactive decay, which makes stockpiling difficult. One recent estimate suggests that global stocks of civilian-use tritium are just 25–30kg, while commissioning and startup of a single commercial fusion reactor may require up to 10kg. The largest source of civilian-use tritium is Canada, which produces ~2kg/yr as a byproduct of heavy water reactor operation, but most of that material is intended to fuel the International Thermonuclear Experimental Reactor (ITER) in the next decade. This tritium production is directly coupled to the power generation rate of its fleet of Canadian Deuterium Uranium (CANDU) reactors; therefore, the only way to increase the tritium production rate is to build more CANDU power reactors.
The National Nuclear Security Administration (NNSA) (an Office of the U.S. Department of Energy (DOE)) – in cooperation with the Tennessee Valley Authority (TVA) – will produce up-to ~4kg of tritium over the next fuel cycles (i.e., ~18-month cycles offset by 6 months) for the two Watts Bar nuclear (WBN) reactors. This would exceed the current, combined 2.8kg production goal, which could be further outstripped if the reactors were operated at their maximum licensed limit, producing ~4.7kg of tritium. All this tritium is designated for military use. However, the NNSA and DOE could leverage production capacities in excess of defense requirements to promote the deployment of FOAK reactors and support U.S. leadership in fusion energy. The DOE could build off the success of its current Milestone-Based Fusion Program by integrating the option for additional tritium availability to meet the commissioning demands of pilot and commercial fusion reactors.
This program could be called “Gigatons-to-Gigawatts” (GtG), a name inspired by one of the most successful fissile material reduction programs in history Megatons-to-Megawatts. The increased scale signifies much higher energy densities contained in tritium vs. the uranium commonly used to fuel fission reactors. Fusion and fission reactor technologies also have very different nonproliferation implications. U.S. national security and nonproliferation goals would be furthered by a systematic transition from fission to fusion energy. Lowering reliance on dual-use nuclear fuel cycle technologies such as centrifuges for uranium enrichment would lower overall proliferation risks. Just as it did by promoting an open fuel cycle, the United States could leverage its technological leadership to promote the adoption of a more proliferation-resistant fusion infrastructure.
However, it is important to note another key difference with Megatons-to-Megawatts: because GtG leverages near-term tritium production capacities in concert with reserves rather than repurposing stockpiled weapons-useable material for civilian use such a program could affect the U.S. nuclear deterrent posture as well. The National Nuclear Security Administration (NNSA) Strategic Integrated Roadmap highlights the goal to “Demonstrate enhanced tritium production capability” for 2025 which is coded as “Nuclear Deterrent.” The anticipated excess production quantities noted above would correspond with this goal. Starting from this demonstrated capability, a GtG program would extend this production capacity into a longer-term effort directed toward a fusion energy future. Furthermore, in support of the long-term goal of nuclear disarmament, GtG would also provide a ready-made framework for repurposing valuable tritium from decommissioned warheads.
One way the United States demonstrates the credibility of its nuclear deterrent is through the Stockpile Stewardship and Management Plan (SSMP). Allies and adversaries alike must believe that the United States has sufficient tritium capability to replenish this critical and slowly decaying resource. An enhanced tritium production capability also has a supporting role to play in reassuring U.S. policymakers that key material design requirements are being sustainably met and that future nuclear weapon tests will be unnecessary. Even though GtG would be programmatically dedicated to the peaceful use of tritium, the technological mechanisms used to reach this goal would nonetheless be compatible with and/or even complementary to the existing nuclear defense posture.
Key facts highlighted in the 2024 Fusion Industry Association (FIA) global reports include: (i) tritium remains the key fuel source for most fusion technologies being developed; (ii) tritium self-sufficiency was seen as one of the major near-term challenges and by a slim margin the major challenge after 2030; and (iii) supply chain partners noted tritium was one of the top 3 constraints to scalability. The easiest reaction to achieve is deuterium–tritium (D–T) fusion. Other more technologically challenging approaches to fusion energy rely on different reactions such as deuterium–deuterium (D–D) and deuterium–Helium-3 (D–He-3) fusion. The Earth has a functionally limitless supply of deuterium; however, even though He-3 is radioactively stable, it slowly leaks from the atmosphere into space. Until humanity can mine the vast quantities of He-3 on the moon, one of the only terrestrial sources of this material is from the tritium decay process. A GtG program would directly support an increase in tritium supply and indirectly support long-term He-3 reserves since it can be stockpiled. Even if fusion with He-3 proves viable, it will be necessary to produce the tritium first.
Once commercial fusion reactors begin operation, breeding tritium to replace burned fuel is a major concern because there is no alternative supply sufficient to replace shortfalls from even modest inefficiency. Operating a 1 GW fusion reactor for a year may require more than 55kg of tritium. Tritium self-sufficiency is nonnegotiable for a functional fusion industry. If technological development falters as companies strive toward a sustainable tritium breeding cycle, they may find themselves in the awkward position of needing tritium more than additional funding.
Of the countries leading the way in private fusion ventures and public investment, the only not closely allied with the U.S. is China, which is also the country most capable of leveraging military tritium production for fusion R&D. In stark contrast with the United States, there is no public information on Chinese tritium production capacities or how much they currently possess. Since China is rapidly expanding their nuclear weapon stockpile, their material margins for repurposing tritium for peaceful-use material will be constrained. If a U.S. investment of tritium into fusion R&D accelerates the growth of domestic companies, then China may be forced to choose between advancing their nuclear weapons agenda and competing with the West for a fusion energy breakthrough.
The United States already has a significant lead in technological capabilities for future generations of fusion energy based on Inertial Confinement Fusion (ICF). The National Ignition Facility (NIF) at Lawrence Livermore National Labs (LLNL) first demonstrated fusion ignition from ICF using tritium in 2022. Largely heralded as a breakthrough for the future of nuclear energy, the facility and ICF tests also provide critical, experimental support for the SSMP. To better position the United States to capitalize on these long-term investments in science and technology, fusion energy leadership should not be ceded to other nations.
Plan of Action
Recommendation 1. Name a White House “Gigatons-to-Gigawatts” czar to coordinate a long-term tritium strategy and interagency cooperation harmonizing national security and fusion energy leadership goals.
A Senior Advisor on the National Security Team of the White House Office of Science and Technology Policy (OSTP) serving as the White House czar for GtG would (i) guide and lead efforts, (ii) coordinate interagency partners, and (iii) facilitate private/public stakeholder forums. Key interagency partners include:
- The Nuclear Weapons Council (NWC)
- DOE National Nuclear Security Administration (NNSA) Office of Tritium and Domestic Uranium Enrichment
- DOE NNSA Tritium Modernization Program (NA-19)
- DOE NNSA Office of Nuclear Material Integration (ONMI) (NA-532)
- DOE Office of Science
- The Office of Fusion Energy Sciences (FES) at DOE Office of Science
- DOE Advanced Research Projects Agency – Energy (ARPA-E)
- State Bureau of International Security and Nonproliferation (ISN)
- TVA Tritium Production Program
- Nuclear Regulatory Commission (NRC) Office of Nuclear Material Safety and Safeguards (NMSS)
- Savannah River National Lab (SRNL)
- Los Alamos National Lab (LANL)
- Pacific Northwest National Lab (PNNL)
- Idaho National Lab (INL) Safety and Tritium Applied Research (STAR)
- Environmental Protection Agency (EPA) Office of Radiation and Indoor Air (ORIA)
- Fusion Energy Sciences Advisory Committee (FESAC)
Key private partners include:
- Savannah River Nuclear Solutions (SRNS) and the Savannah River Tritium Enterprise (SRTE) program
- Westinghouse Government Services (WGS) Columbia Fuel Fabrication Facility (CFFF)
- Fusion Industry Association (FIA)
A central task of the GtG czar would be to coordinate with the NWC to review Presidential Policy Directive 9 (PPD-9) and associated/superseding planning documents related to the assessment of tritium demand requirements including (i) laboratory research, development, and surveillance and (ii) presidentially mandated tritium reserve. These two components of the tritium requirement could potentially be expanded to address GtG needs. If deemed appropriate, the President of the United States could be advised to expand the presidentially mandated reserve. Otherwise, the former requirement could be expanded based on optimal quantities to stand up a GtG program capability. A reference target would be the accumulation of ~10kg of tritium on projected timelines for commissioning full-scale FOAK fusion reactors.
The following recommendations could be coordinated by a GtG czar or done independently.
Recommendation 2. The Secretary of Energy should direct the Office of Science to evaluate the Milestone-Based Fusion Development Program for integrating GtG tritium production and supply targets with projected industry demands for commissioning fusion power plants.
The Milestone-Based Fusion Development Program has already provided awards of $46 million to 8 US companies. It is crucial to ensure that any tritium produced for a GtG program is not accumulated without a viable success path for FOAK fusion plant commissioning. Given the modest production capacities currently available at the WBN site, timelines of 5–10 years will be necessary to accumulate tritium. Each fuel cycle could allow for adjustments in production targets, but sufficient lead time will be required to anticipate and plan for necessary core changes and fuel-assembly production.
GtG tritium awards aligned with the Milestone-Based Fusion Development Program would also be more viable and attractive if costs were equitably shared between private awardees and the DOE. The U.S. Government produces tritium at WBN at a premium of ~$50,000/g whereas the market rate for tritium produced in Canada is closer to $30,000/g. A fusion company awarded tritium through the GtG program should be required to pay the prevailing market rate for tritium upon extraction at the Savannah River Site (SRS). This would allow a fusion company to benefit from increased tritium availability, while the DOE shoulders the cost differences of Tritium-Producing Burnable Absorber Rod (TPBAR) production methods. Additionally, this pay-as-you-go requirement will incentivize fusion energy companies to lay out realistic timeframes for FOAK reactor deployments.
The Director of the Office of Science should also direct the FESAC to prepare a report on tritium demand scenarios that would apply to leading fusion technology development timelines and assess the necessary tritium breeding efficiencies needed to sustain fusion power plant operations. The FESAC should give special consideration to projecting possible mitigation and recovery strategies for tritium breeding shortfalls. The committee should also provide thresholds for FOAK fusion reactors’ short-term recoverability from tritium breeding shortfalls. Tritium quantities based on this FESAC report should be considered for future tritium hedges after these fusion reactors begin power operations.
Recommendation 3. The NNSA ONMI (NA-532) should coordinate an interagency review of the tritium supply chain infrastructure.
Raising tritium production targets beyond previously projected requirements would necessitate review from TPBAR assembly at Westinghouse’s CFFF, irradiation at TVA’s Watts Bar Reactors, and then extraction and processing through the SRTE program at SRS. Because this review naturally involves civilian reactors and the transport of nuclear materials the NRC should also be consulted to ensure regulatory compliance is maintained. This review will provide realistic bounding limits to the quantities of tritium and production timelines that could be designated for a GtG program. The outcome of this review will inform industry-facing efforts to better assess how additional tritium supplies could best support fusion energy R&D and pilot plant commissioning.
As part of this process, the NA-532 office should determine which existing tritium supply chain models are best suited for assessing commercial applications, including the LANL Tritium Supply and Demand Model and those developed internally by the NNSA. If no model is determined fit for purpose, then a new model should be developed to best capture the dynamics of commercial fusion R&D. In any case, existing models should form the basis for integrating military requirements and civilian markets to ensure a GtG program adequately accounts for both.
An added-value option for this recommendation would be to prepare an unclassified and publicly accessible version of the commercial tritium supply chain model. This would reinforce the transparency and public accountability already built into the production of tritium in the commercial power reactors at Watts Bar. Furthermore, such a resource would also help explain the rationale and intent behind the use of public funds to support fusion R&D and the commissioning of FOAK fusion reactors.
Recommendation 4. The Secretary of Energy should direct a review of DOE Technical Standards for addressing tritium-related radiological risks.
While the general scientific consensus is that low-level tritium exposure poses negligible human health and ecosystem risks, there are several unknowns that should be better understood before the advent of fusion energy releases unprecedented quantities of tritium into the environment. This adequacy review should include at least [i] a comprehensive analysis of risks from Organically Bound Tritium (OBT) and [ii] more precisely quantifying and considering the potential for damaging mitochondrial DNA and fetuses. These efforts would help ensure the responsible, consent-based rollout of tritium-intensive technologies and allow for an informed public to better understand the magnitude of risks to be weighed against potential benefits.
Key DOE Technical Standards to include in this review:
- Derived Concentration Technical Standard (DOE-STD-1196-2022)
- Internal Dosimetry (DOE-STD-1121-2008 (Reaffirmed 2022))
- Nuclear Materials Control and Accountability (DOE-STD-1194-2019)
Recommendation 5. The Administrator of the Environmental Protection Agency (EPA) should direct the Office of Radiation and Indoor Air (ORIA) to assess the adequacy of radioactive dose calculations in the Federal Guidance Report on External Exposure to Radionuclides in Air, Water, and Soil (FGR 15) last issued in 2019.
This recommendation, along with recommendation 3 above, will provide sufficient lead time to address any uncertainties and unknowns regarding the radiological risks posed by tritium. As in this previous case, this adequacy review should include at least [i] a comprehensive analysis of risks from Organically Bound Tritium (OBT) and [ii] more precisely quantifying and considering the potential for damaging mitochondrial DNA and fetuses. FGR 15 currently calculates effective dose rates for “computational phantom” models of 6 different age groups, including newborns, that incorporate both male and female sex-specific tissues. However, effective dose rates and potential effects are not considered for developing fetuses. The uncertainty surrounding tritium’s radiological risks prompts an extensive precautionary approach to potential exposures for declared pregnant workers. However, the potential for higher levels of tritium exposure for pregnant members of the public should also be taken into consideration when assessing the radiological risks of fusion energy.
Conclusion
With a strategically calibrated GtG program, the United States could remain technology leaders in fusion energy and potentially reduce the rollout timeline of a multi-unit fleet by several years. In the context of state-level technological competition and a multi-polar nuclear security environment, years matter. A strategic GtG reserve will take years to plan and accumulate to ensure sufficient tritium is available at the right time.
The long-term utility of a GtG framework is not limited to the designation of new tritium production for peaceful use. Once nuclear-weapons states return to the negotiating table to reduce the number of nuclear weapons in the world, the United States would have a clear roadmap for repurposing the tritium from decommissioned weapons in support of fusion power. Previously, the United States held onto large reserves of this valuable and critical material for years while transitioning from military to civilian production. The years between 2025 and 2040 will provide more chances to put that material to productive use for fusion energy. Let us not waste this opportunity to ensure the U.S. remains at the vanguard of the fusion revolution.
This action-ready policy memo is part of Day One 2025 — our effort to bring forward bold policy ideas, grounded in science and evidence, that can tackle the country’s biggest challenges and bring us closer to the prosperous, equitable and safe future that we all hope for whoever takes office in 2025 and beyond.
PLEASE NOTE (February 2025): Since publication several government websites have been taken offline. We apologize for any broken links to once accessible public data.
A U.S. Government Accountability Office (GAO) report from 2000 provided unclassified approximations of total life-cycle cost ranged from ~$34,000 to $57,000 per gram of tritium. With several program delays and at least one major capital investment (i.e., a 500,000 gallon Tritiated Water Storage Tank (TWST) system) costing ~$20 million, the actual life-cycle costs are likely higher. The cost of tritium produced in Canada is closer to $30,000 per gram, but, as noted above, only fixed and limited amounts of tritium can be made available through this process.
This is unlikely. The SSMP projects tritium needs far enough into the future that demand changes could allow for adjustments to production levels over the span of 1–2 fuel cycles (i.e., one and a half to three years). Barring a catastrophic loss of military tritium reserves or a significant nuclear accident at Watts Bar, there is unlikely to be a tritium supply emergency requiring an immediate response.
Historical tritium production amounts and capacities at SRS remain restricted data. However, due to NRC regulatory requirements for commercial reactors, this information cannot be protected for tritium production at Watts Bar. Since tritium production transparency has been the norm since 2003, the United States may further demonstrate nuclear stockpile credibility by openly producing material in excess of current military requirements.
Unobligated fuel demand would slightly increase. Unobligated fuel requirements are largely a sunk cost. Regardless of how many TPBARs are being irradiated the entire core will be composed of unobligated fuel. However, increased tritium production (i.e., irradiating more TPBARs) would require additional fresh fuel bundles per fuel cycle. The 2024 SSMP currently projects meeting Watts Bar’s unobligated fuel needs through 2044.
This would possibly require new license amendments for each reactor, but if the amounts were below the previously analyzed conditions, then a new Environmental Impact Statement (EIS) would not be required. The current license for each reactor allows for the irradiation of up to 2,496 TPBARs per fuel cycle per reactor. The EIS analysis is bounded at a maximum of 6,000 TPBARs combined per fuel cycle. The average yield of each TPBAR is 0.95g of tritium.
Fusion industry leaders have demonstrated confidence that existing and future supplies of civilian-use tritium, while modest, are sufficient to fuel the necessary near-term R&D. In particular, the planned refurbishments to aging Canadian CANDU reactors and the additional delays at ITER have propped open the tritium window for several more years until tritium breeding blanket technologies can mature. However, tritium supply chain bottlenecks could constrain industry momentum and/or advantage states capable of backstopping any shortages.
Policy Experiment Stations to Accelerate State and Local Government Innovation
The federal government transfers approximately $1.1 trillion dollars every year to state and local governments. Yet most states and localities are not evaluating whether the programs deploying these funds are increasing community well-being. Similarly, achieving important national goals like increasing clean energy production and transmission often requires not only congressional but also state and local policy reform. Yet many states and localities are not implementing the evidence-based policy reforms necessary to achieve these goals.
State and local government innovation is a problem not only of politics but also of capacity. State and local governments generally lack the technical capacity to conduct rigorous evaluations of the efficacy of their programs, search for reliable evidence about programs evaluated in other contexts, and implement the evidence-based programs with the highest chances of improving outcomes in their jurisdictions. This lack of capacity severely constrains the ability of state and local governments to use federal funds effectively and to adopt more effective ways of delivering important public goods and services. To date, efforts to increase the use of evaluation evidence in federal agencies (including the passage of the Evidence Act) have not meaningfully supported the production and use of evidence by state and local governments.
Despite an emerging awareness of the importance of state and local government innovation capacity, there is a shortage of plausible strategies to build that capacity. In the words of journalist Ezra Klein, we spend “too much time and energy imagining the policies that a capable government could execute and not nearly enough time imagining how to make a government capable of executing them.”
Yet an emerging body of research is revealing that an effective strategy to build government innovation capacity is to partner government agencies with local universities on scientifically rigorous evaluations of the efficacy of their programs, curated syntheses of reliable evaluation evidence from other contexts, and implementation of evidence-based programs with the best chances of success. Leveraging these findings, along with recent evidence of the striking efficacy of the national network of university-based “Agriculture Experiment Stations” established by the Hatch Act of 1887, we propose a national network of university-based “Policy Experiment Stations” or policy innovation labs in each state, supported by continuing federal and state appropriations and tasked with accelerating state and local government innovation.
Challenge
Advocates of abundance have identified “failed public policy” as an increasingly significant barrier to economic growth and community flourishing. Of particular concern are state and local policies and programs, including those powered by federal funds, that do not effectively deliver critically important public goods and services like health, education, safety, clean air and water, and growth-oriented infrastructure.
Part of the challenge is that state and local governments lack capacity to conduct rigorous evaluations of the efficacy of their policies and programs. For example, the American Rescue Plan, the largest one-time federal investment in state and local governments in the last century, provided $350 billion in State and Local Fiscal Recovery Funds to state, territorial, local, and Tribal governments to accelerate post-pandemic economic recovery. Yet very few of those investments are being evaluated for efficacy. In a recent survey of state policymakers, 59% of those surveyed cited “lack of time for rigorous evaluations” as a key obstacle to innovation. State and local governments also typically lack the time, resources, and technical capacity to canvass evaluation evidence from other settings and assess whether a program proven to improve outcomes elsewhere might also improve outcomes locally. Finally, state and local governments often don’t adopt more effective programs even when they have rigorous evidence that these programs are more effective than the status quo, because implementing new programs disrupts existing workflows.
If state and local policymakers don’t know what works and what doesn’t, and/or aren’t able to overcome even relatively minor implementation challenges when they do know what works, they won’t be able to spend federal dollars more effectively, or more generally to deliver critical public goods and services.
Opportunity
A growing body of research on government innovation is documenting factors that reliably increase the likelihood that governments will implement evidence-based policy reform. First, government decision makers are more likely to adopt evidence-based policy reforms when they are grounded in local evidence and/or recommended by local researchers. Boston-based researchers sharing a Boston-based study showing that relaxing density restrictions reduces rents and house prices will do less to convince San Francisco decision makers than either a San Francisco-based study, or San Francisco-based researchers endorsing the evidence from Boston. Proximity matters for government innovation.
Second, government decision makers are more likely to adopt evidence-based policy reforms when they are engaged as partners in the research projects that produce the evidence of efficacy, helping to define the set of feasible policy alternatives and design new policy interventions. Research partnerships matter for government innovation.
Third, evidence-based policies are significantly more likely to be adopted when the policy innovation is part of an existing implementation infrastructure, or when agencies receive dedicated implementation support. This means that moving beyond incremental policy reforms will require that state and local governments receive more technical support in overcoming implementation challenges. Implementation matters for government innovation.
We know that the implementation of evidence-based policy reform produces returns for communities that have been estimated to be on the order of 17:1. Our partners in government have voiced their direct experience of these returns. In Puerto Rico, for example, decision makers in the Department of Education have attributed the success of evidence-based efforts to help students learn to the “constant communication and effective collaboration” with researchers who possessed a “strong understanding of the culture and social behavior of the government and people of Puerto Rico.” Carrie S. Cihak, the evidence and impact officer for King County, Washington, likewise observes,
“It is critical to understand whether the programs we’re implementing are actually making a difference in the communities we serve. Throughout my career in King County, I’ve worked with County teams and researchers on evaluations across multiple policy areas, including transportation access, housing stability, and climate change. Working in close partnership with researchers has guided our policymaking related to individual projects, identified the next set of questions for continual learning, and has enabled us to better apply existing knowledge from other contexts to our own. In this work, it is essential to have researchers who are committed to valuing local knowledge and experience–including that of the community and government staff–as a central part of their research, and who are committed to supporting us in getting better outcomes for our communities.”
The emerging body of evidence on the determinants of government innovation can help us define a plan of action that galvanizes the state and local government innovation necessary to accelerate regional economic growth and community flourishing.
Plan of Action
An evidence-based plan to increase state and local government innovation needs to facilitate and sustain durable partnerships between state and local governments and neighboring universities to produce scientifically rigorous policy evaluations, adapt evaluation evidence from other contexts, and develop effective implementation strategies. Over a century ago, the Hatch Act of 1887 created a remarkably effective and durable R&D infrastructure aimed at agricultural innovation, establishing university-based Agricultural Experiment Stations (AES) in each state tasked with developing, testing, and translating innovations designed to increase agricultural productivity.
Locating university-based AES in every state ensured the production and implementation of locally-relevant evidence by researchers working in partnership with local stakeholders. Federal oversight of the state AES by an Office of Experiment Stations in the US Department of Agriculture ensured that work was conducted with scientific rigor and that local evidence was shared across sites. Finally, providing stable annual federal appropriations for the AES, with required matching state appropriations, ensured the durability and financial sustainability of the R&D infrastructure. This infrastructure worked: agricultural productivity near the experiment stations increased by 6% after the stations were established.
Congress should develop new legislation to create and fund a network of state-based “Policy Experiment Stations.”
The 119th Congress that will convene on January 3, 2025 can adapt the core elements of the proven-effective network of state-based Agricultural Experiment Stations to accelerate state and local government innovation. Mimicking the structure of 7 USC 14, federal grants to states would support university-based “Policy Experiment Stations” or policy innovation labs in each state, tasked with partnering with state and local governments on (1) scientifically rigorous evaluations of the efficacy of state and local policies and programs; (2) translations of evaluation evidence from other settings; and (3) overcoming implementation challenges.
As in 7 USC 14, grants to support state policy innovation labs would be overseen by a federal office charged with ensuring that work was conducted with scientific rigor and that local evidence was shared across sites. We see two potential paths for this oversight function, paths that in turn would influence legislative strategy.
Pathway 1: This oversight function could be located in the Office of Evaluation Sciences (OES) in the General Services Administration (GSA). In this case, the congressional committees overseeing GSA, namely the House Committee on Oversight and Responsibility and the Senate Committee on Homeland Security and Governmental Affairs, would craft legislation providing for an appropriation to GSA to support a new OES grants program for university-based policy innovation labs in each state. The advantage of this structure is that OES is a highly respected locus of program and policy evaluation expertise.
Pathway 2: Oversight could instead be located in the Directorate of Technology, Innovation, and Partnerships in the National Science Foundation (NSF TIP). In this case, the House Committee on Science, Space, and Technology and the Senate Committee on Commerce, Science, and Transportation would craft legislation providing for a new grants program within NSF TIP to support university-based policy innovation labs in each state. The advantage of this structure is that NSF is a highly respected grant-making agency.
Either of these paths is feasible with bipartisan political will. Alternatively, there are unilateral steps that could be taken by the incoming administration to advance state and local government innovation. For example, the Office of Management and Budget (OMB) recently released updated Uniform Grants Guidance clarifying that federal grants may be used to support recipients’ evaluation costs, including “conducting evaluations, sharing evaluation results, and other personnel or materials costs related to the effective building and use of evidence and evaluation for program design, administration, or improvement.” The Uniform Grants Guidance also requires federal agencies to assess the performance of grant recipients, and further allows federal agencies to require that recipients use federal grant funds to conduct program evaluations. The incoming administration could further update the Uniform Grants Guidance to direct federal agencies to require that state and local government grant recipients set aside grant funds for impact evaluations of the efficacy of any programs supported by federal funds, and further clarify the allowability of subgrants to universities to support these impact evaluations.
Conclusion
Establishing a national network of university-based “Policy Experiment Stations” or policy innovation labs in each state, supported by continuing federal and state appropriations, is an evidence-based plan to facilitate abundance-oriented state and local government innovation. We already have impressive examples of what these policy labs might be able to accomplish. At MIT’s Abdul Latif Jameel Poverty Action Lab North America, the University of Chicago’s Crime Lab and Education Lab, the University of California’s California Policy Lab, and Harvard University’s The People Lab, to name just a few, leading researchers partner with state and local governments on scientifically rigorous evaluations of the efficacy of public policies and programs, the translation of evidence from other settings, and overcoming implementation challenges, leading in several cases to evidence-based policy reform. Yet effective as these initiatives are, they are largely supported by philanthropic funds, an infeasible strategy for national scaling.
In recent years we’ve made massive investments in communities through federal grants to state and local governments. We’ve also initiated ambitious efforts at growth-oriented regulatory reform which require not only federal but also state and local action. Now it’s time to invest in building state and local capacity to deploy federal investments effectively and to galvanize regional economic growth. Emerging research findings about the determinants of government innovation, and about the efficacy of the R&D infrastructure for agricultural innovation established over a century ago, give us an evidence-based roadmap for state and local government innovation.
This action-ready policy memo is part of Day One 2025 — our effort to bring forward bold policy ideas, grounded in science and evidence, that can tackle the country’s biggest challenges and bring us closer to the prosperous, equitable and safe future that we all hope for whoever takes office in 2025 and beyond.
PLEASE NOTE (February 2025): Since publication several government websites have been taken offline. We apologize for any broken links to once accessible public data.
Accelerating Materials Science with AI and Robotics
Innovations in materials science enable innumerable downstream innovations: steel enabled skyscrapers, and novel configurations of silicon enabled microelectronics. Yet progress in materials science has slowed in recent years. Fundamentally, this is because there is a vast universe of potential materials, and the only way to discover which among them are most useful is to experiment. Today, those experiments are largely conducted by hand. Innovations in artificial intelligence and robotics will allow us to accelerate the search process using foundation AI models for science research and automate much of the experimentation with robotic, self-driving labs. This policy memo recommends the Department of Energy (DOE) lead this effort because of its unique expertise in supercomputing, AI, and its large network of National Labs.
Challenge and Opportunity
Take a look at your smartphone. How long does its battery last? How durable is its frame? How tough is its screen? How fast and efficient are the chips inside it?
Each of these questions implicates materials science in fundamental ways. The limits of our technological capabilities are defined by the limits of what we can build, and what we can build is defined by what materials we have at our disposal. The early eras of human history are named for materials: the Stone Age, the Bronze Age, the Iron Age. Even today, the cradle of American innovation is Silicon Valley, a reminder that even our digital era is enabled by finding innovative ways to assemble matter to accomplish novel things.
Materials science has been a driver of economic growth and innovation for decades. Improvements to silicon purification and processing—painstakingly worked on in labs for decades—fundamentally enabled silicon-based semiconductors, a $600 billion industry today that McKinsey recently projected would double in size by 2030. The entire digital economy, conservatively estimated by the Bureau of Economic Analysis (BEA) at $3.7 trillion in the U.S. alone, in turn, rests on semiconductors. Plastics, another profound materials science innovation, are estimated to have generated more than $500 billion in economic value in the U.S. last year. The quantitative benefits are staggering, but even qualitatively, it is impossible to imagine modern life without these materials.
However, present-day materials are beginning to show their age. We need better batteries to accelerate the transition to clean energy. We may be approaching the limits of traditional methods of manufacturing semiconductors in the next decade. We require exotic new forms of magnets to bring technologies like nuclear fusion to life. We need materials with better thermal properties to improve spacecraft.
Yet materials science and engineering—the disciplines of discovering and learning to use new materials—have slowed down in recent decades. The low-hanging fruit has been plucked, and the easy discoveries are old news. We’re approaching the limits of what our materials can do because we are also approaching the limits of what the traditional practice of materials science can do.
Today, materials science proceeds at much the same pace as it did half a century ago: manually, with small academic labs and graduate students formulating potential new combinations of elements, synthesizing those combinations, and studying their characteristics. Because there are more ways to configure matter than there are atoms in the universe, manually searching through the space of possible materials is an impossible task.
Fortunately, AI and robotics present an opportunity to automate that process. AI foundation models for physics and chemistry can be used to simulate potential materials with unprecedented speed and low cost compared to traditional ab initio methods. Robotic labs (also known as “self-driving labs”) can automate the manual process of performing experiments, allowing scientists to synthesize, validate, and characterize new materials twenty-four hours a day at dramatically lower costs. The experiments will generate valuable data for further refining the foundation models, resulting in a positive feedback loop. AI language models like OpenAI’s GPT-4 can write summaries of experimental results and even help ideate new experiments. The scientists and their grad students, freed from this manual and often tedious labor, can do what humans do best: think creatively and imaginatively.
Achieving this goal will require a coordinated effort, significant investment, and expertise at the frontiers of science and engineering. Because much of materials science is basic R&D—too far from commercialization to attract private investment—there is a unique opportunity for the federal government to lead the way. As with much scientific R&D, the economic benefits of new materials science discoveries may take time to emerge. One literature review estimated that it can take roughly 20 years for basic research to translate to economic growth. Research indicates that the returns—once they materialize—are significant. A study from the Federal Reserve Bank of Dallas suggests a return of 150-300% on federal R&D spending.
The best-positioned department within the federal government to coordinate this effort is the DOE, which has many of the key ingredients in place: a demonstrated track record of building and maintaining the supercomputing facilities required to make physics-based AI models, unparalleled scientific datasets with which to train those models collected over decades of work by national labs and other DOE facilities, and a skilled scientific and engineering workforce capable of bringing challenging projects to fruition.
Plan of Action
Achieving the goal of using AI and robotics to simulate potential materials with unprecedented speed and low cost, and benefit from the discoveries, rests on five key pillars:
- Creating large physics and chemistry datasets for foundation model training (estimated cost: $100 million)
- Developing foundation AI models for materials science discovery, either independently or in collaboration with the private sector (estimated cost: $10-100 million, depending on the nature of the collaboration);
- Building 1-2 pilot self-driving labs (SDLs) aimed at establishing best practices, building a supply chain for robotics and other equipment, and validating the scientific merit of SDLs (estimated cost: $20-40 million);
- Making self-driving labs an official priority of the DOE’s preexisting FASST initiative (described below);
- Directing the DOE’s new Foundation for Energy Security and Innovation (FESI) to prioritize establishing fellowships and public-private partnerships to support items (1) and (2), both financially and with human capital.
The total cost of the proposal, then, is estimated at between $130-240 million. The potential return on this investment, though, is far higher. Moderate improvements to battery materials could drive tens or hundreds of billions of dollars in value. Discovery of a “holy grail” material, such as a room-temperature, ambient-pressure superconductor, could create trillions of dollars in value.
Creating Materials Science Foundation Model Datasets
Before a large materials science foundation model can be trained, vast datasets must be assembled. DOE, through its large network of scientific facilities including particle colliders, observatories, supercomputers, and other experimental sites, collects enormous quantities of data–but this, unfortunately, is only the beginning. DOE’s data infrastructure is out-of-date and fragmented between different user facilities. Data access and retention policies make sharing and combining different datasets difficult or impossible.
All of these policy and infrastructural decisions were made far before training large-scale foundation models was a priority. They will have to be changed to capitalize on the newfound opportunity of AI. Existing DOE data will have to be reorganized into formats and within technical infrastructure suited to training foundation models. In some cases, data access and retention policies will need to be relaxed or otherwise modified.
In other cases, however, highly sensitive data will need to be integrated in more sophisticated ways. A 2023 DOE report, recognizing the problems with DOE data infrastructure, suggests developing federated learning capabilities–an active area of research in the broader machine learning community–which would allow for data to be used for training without being shared. This would, the report argues, ”allow access and connections to the information through access control processes that are developed explicitly for multilevel privacy.”
This work will require deep collaboration between data scientists, machine learning scientists and engineers, and domain-specific scientists. It is, by far, the least glamorous part of the process–yet it is the necessary groundwork for all progress to follow.
Building AI Foundation Models for Science
Fundamentally, AI is a sophisticated form of statistics. Deep learning, the broad approach that has undergirded all advances in AI over the past decade, allows AI models to uncover deep patterns in extremely complex datasets, such as all the content on the internet, the genomes of millions of organisms, or the structures of thousands of proteins and other biomolecules. Models of this kind are sometimes loosely referred to as “foundation models.”
Foundation models for materials science can take many different forms, incorporating various aspects of physics, chemistry, and even—for the emerging field of biomaterials—biology. Broadly speaking, foundation models can help materials science in two ways: inverse design and property prediction. Inverse design allows scientists to input a given set of desired characteristics (toughness, brittleness, heat resistance, electrical conductivity, etc.) and receive a prediction for what material might be able to achieve those properties. Property prediction is the opposite flow of information, inputting a given material and receiving a prediction of what properties it will have in the real world.
DOE has already proposed creating AI foundation models for materials science as part of its Frontiers in Artificial Intelligence for Science, Security and Technology (FASST) initiative. While this initiative contains numerous other AI-related science and technology objectives, supporting it would enable the creation of new foundation models, which can in turn be used to support the broader materials science work.
DOE’s long history of stewarding America’s national labs makes it the best-suited home for this proposal. DOE labs and other DOE sub-agencies have decades of data from particle accelerators, nuclear fusion reactors, and other specialized equipment rarely seen in other facilities. These labs have performed hundreds of thousands of experiments in physics and chemistry over their lifetimes, and over time, DOE has created standardized data collection practices. AI models are defined by the data that they are trained with, and DOE has some of the most comprehensive physics and chemistry datasets in the country—if not the world.
The foundation models created by DOE should be made available to scientists. The extent of that availability should be determined by the sensitivity of the data used to train the model and other potential risks associated with broad availability. If, for example, a model was created using purely internal or otherwise sensitive DOE datasets, it might have to be made available only to select audiences with usage monitored; otherwise, there is a risk of exfiltrating sensitive training data. If there are no such data security concerns, DOE could choose to fully open source the models, meaning their weights and code would be available to the general public. Regardless of how the models themselves are distributed, the fruits of all research enabled by both DOE foundation models and self-driving labs should be made available to the academic community and broader public.
Scaling Self-Driving Labs
Self-driving labs are largely automated facilities that allow robotic equipment to autonomously conduct scientific experiments with human supervision. They are well-suited to relatively simple, routine experiments—the exact kind involved in much of materials science. Recent advancements in robotics have been driven by a combination of cheaper hardware and enhanced AI models. While fully autonomous humanoid robots capable of automating arbitrary manual labor are likely years away, it is now possible to configure facilities to automate a broad range of scripted tasks.
Many experiments in materials science involve making iterative tweaks to variables within the same broad experimental design. For example, a grad student might tweak the ratios of the elements that constitute the material, or change the temperature at which the elements are combined. These are highly automatable tasks. Furthermore, by allowing multiple experiments to be conducted in parallel, self-driving labs allow scientists to rapidly accelerate the pace at which they conduct their work.
Creating a successful large-scale self-driving lab will require collaboration with private sector partners, particularly robot manufacturers and the creators of AI models for robotics. Fortunately, the United States has many such firms. Therefore, DOE should initiate a competitive bidding process for the robotic equipment that will be housed within its self-driving labs. Because DOE has experience in building lab facilities, it should directly oversee the construction of the self-driving lab itself.
The United States already has several small-scale self-driving labs, primarily led by investments at DOE National Labs. The small size of these projects, however, makes it difficult to achieve the economies of scale that are necessary for self-driving labs to become an enduring part of America’s scientific ecosystem.
AI creates additional opportunities to expand automated materials science. Frontier language and multi-modal models, such as OpenAI’s GPT-4o, Anthropic’s Claude 3.5, and Google’s Gemini family, have already been used to ideate scientific experiments, including directing a robotic lab in the fully autonomous synthesis of a known chemical compound. These models would not operate with full autonomy. Instead, scientists would direct the inquiry and the design of the experiment, with the models autonomously suggesting variables to tweak.
Modern frontier models have substantial knowledge in all fields of science, and can hold all of the academic literature relevant to a specific niche of materials science within their active attention. This combination means that they have—when paired with a trained human—the scientific intuition to iteratively tweak an experimental design. They can also write the code necessary to direct the robots in the self-driving lab. Finally, they can write summaries of the experimental results—including the failures. This is crucial, because, given the constraints on their time, scientists today often only report their successes in published writing. Yet failures are just as important to document publicly to avoid other scientists duplicating their efforts.
Once constructed, this self-driving lab infrastructure can be a resource made available as another DOE user facility to materials scientists across the country, much as DOE supercomputers are today. DOE already has a robust process and infrastructure in place to share in-demand resources among different scientists, again underscoring why the Department is well-positioned to lead this endeavor.
Conclusion
Taken together, materials science faces a grand challenge, yet an even grander opportunity. Room-temperature, ambient-pressure superconductors—permitted by the laws of physics but as-yet undiscovered—could transform consumer electronics, clean energy, transportation, and even space travel. New forms of magnets could enable a wide range of cutting-edge technologies, such as nuclear fusion reactors. High-performance ceramics could improve reusable rockets and hypersonic aircraft. The opportunities are limitless.
With a coordinated effort led by DOE, the federal government can demonstrate to Americans that scientific innovation and technological progress can still deliver profound improvements to daily life. It can pave the way for a new approach to science firmly rooted in modern technology, creating an example for other areas of science to follow. Perhaps most importantly, it can make Americans excited about the future—something that has been sorely lacking in American society in recent decades.
AI is a radically transformative technology. Contemplating that transformation in the abstract almost inevitably leads to anxiety and fear. There are legislative proposals, white papers, speeches, blog posts, and tweets about using AI to positive ends. Yet merely talking about positive uses of AI is insufficient: the technology is ready, and the opportunities are there. Now is the time to act.
This action-ready policy memo is part of Day One 2025 — our effort to bring forward bold policy ideas, grounded in science and evidence, that can tackle the country’s biggest challenges and bring us closer to the prosperous, equitable and safe future that we all hope for whoever takes office in 2025 and beyond.
PLEASE NOTE (February 2025): Since publication several government websites have been taken offline. We apologize for any broken links to once accessible public data.
Compared to “cloud labs” for biology and chemistry, the risks associated with self-driving labs for materials science are low. In a cloud lab equipped with nucleic acid synthesis machines, for example, genetic sequences need to be screened carefully to ensure that they are not dangerous pathogens—a nontrivial task. There are not analogous risks for most materials science applications.
However, given the dual-use nature of many novel materials, any self-driving lab would need to have strong cybersecurity and intellectual property protections. Scientists using self-driving lab facilities would need to be carefully screened by DOE—fortunately, this is an infrastructure DOE possesses already for determining access to its supercomputing facilities.
Not all materials involve easily repeatable, and hence automatable, experiments for synthesis and characterization. But many important classes of materials do, including:
- Thin films and coatings
- Photonic and optoelectronic materials such as perovskites (used for solar panels)
- Polymers and monomers
- Battery and energy storage materials
Over time, additional classes of materials can be added.
DOE can and should be creative and resourceful in finding additional resources beyond public funding for this project. Collaborations on both foundation AI models and scaling self-driving labs between DOE and private sector AI firms can be uniquely facilitated by DOE’s new Foundation for Energy Security and Innovation (FESI), a private foundation created by DOE to support scientific fellowships, public-private partnerships, and other key mission-related initiatives.
Yes. Some private firms have recently demonstrated the promise. In late 2023, Google DeepMind unveiled GNoME, a materials science model that identified thousands of new potential materials (though they need to be experimentally validated). Microsoft’s GenMatter model pushed in a similar direction. Both models were developed in collaboration with DOE National Labs (Lawrence Berkeley in the case of DeepMind, and Pacific Northwest in the case of Microsoft).
Promoting American Resilience Through a Strategic Investment Fund
Critical minerals, robotics, advanced energy systems, quantum computing, biotechnology, shipbuilding, and space are some of the resources and technologies that will define the economic and security climate of the 21st century. However, the United States is at risk of losing its edge in these technologies of the future. For instance, China processes the vast majority of the world’s batteries and critical metals and has successfully launched a quantum communications satellite. The implications are enormous: the U.S. relies on its qualitative technological edge to fuel productivity growth, improve living standards, and maintain the existing global order. Indeed, the Inflation Reduction Act (IRA) and CHIPS Act were largely reactionary moves to shore up atrophied manufacturing capabilities in the American battery and semiconductor industries, requiring hundreds of billions in outlays to catch up. In an ideal world, critical industries would be sufficiently funded well in advance to avoid economically costly catch-up spending.
However, many of these technologies are characterized by long timelines, significant capital expenditures, and low and uncertain profit margins, presenting major challenges for private-sector investors who are required by their limited partners (capital providers such as pension funds, university endowments, and insurance companies) to underwrite to a certain risk-adjusted return threshold. This stands in contrast to technologies like artificial intelligence and pharmaceuticals: While both are also characterized by large upfront investments and lengthy research and development timelines, the financial payoffs are far clearer, incentivizing private sectors to play a leading role in commercialization. This issue for technologies in economically and geopolitically vital industries such as lithium processing and chips is most acute in the “valley of death,” when companies require scale-up capital for an early commercialization effort: the capital required is too large for traditional venture capital, yet too risky for traditional project finance.
The United States needs a strategic investment fund (SIF) to shepherd promising technologies in nationally vital sectors through the valley of death. An American SIF is not intended to provide subsidies, pick political winners or losers, or subvert the role of private capital markets. On the contrary, its role would be to “crowd in” capital by uniquely managing risk that no private or philanthropic entities have the capacity to do. In doing so, an SIF would ensure that the U.S. maintains an edge in critical technologies, promoting economic dynamism and national security in an agile, cost-efficient manner.
Challenges
The Need for Private Investment
A handful of resources and technologies, some of which have yet to be fully characterized, have the potential to play an outsized role in the future economy. Most of these key technologies have meaningful national security implications.
Since ChatGPT’s release in November 2022, artificial intelligence (AI) has experienced a commercial renaissance that has captured the public’s imagination and huge sums of venture dollars, as evidenced by OpenAI’s October 2024 $6.5 billion round at a $150 billion pre-money valuation. However, AI is not the only critical resource or technology that will power the future economy, and many of those critical resources and technologies may struggle to attract the same level of private investment. Consider the following:
- To meet climate goals, the world needs to increase production of lithium by nearly 475%, rare earths by 100%, and nickel by 60% through 2035. For defense applications, rare earths are especially important; the construction of one F-35, for instance, uses 920 pounds of rare earth materials.
- The conflict in Ukraine has unequivocally demonstrated the value of low-cost drones on the battlefield. However, drones also have significant commercial applications, including safety and last-mile delivery. Reducing production and component costs could make a meaningful difference.
- Quantum technology has the potential to exponentially expand compute power, which can be used to simulate biological pathways, accelerate materials development, and process vast amounts of financial data. However, quantum technology can also be used to break existing encryption technologies and safeguard communications. China launched its first quantum communications satellite in 2020.
Few sectors receive the level of consistent venture attention that software technology, most recently in AI, has gotten in the last 18 months. However, this does not make them unbackable or unimportant; on the contrary, technologies that increase mineral recovery yields or make drone engines cheaper should receive sufficient support to get to scale. While private-sector capital markets have supported the development of many important industries, they are not perfect and may miss important opportunities due to information asymmetries and externalities.
Overcoming the Valley of Death
Many strategically important technologies are characterized by high upfront costs and low or uncertain margins, which tends to dissuade investment by private-sector organizations at key inflection points, namely, the “valley of death.”
By their nature, innovative technologies are complex and highly uncertain. However, some factors make future economic value—and therefore financeability—more difficult to ascertain than others. For example, innovative battery technologies that enable long-term storage of energy generated from renewables would greatly improve the economics of utility-scale solar and wind projects. However, this requires production at scale in the face of potential competition from low-cost incumbents. In addition, there is the element of scientific risk itself, as well as the question of customer adoption and integration. There are many good reasons why technologies and companies that seem feasible, economical, and societally valuable do not succeed.
These dynamics result in lopsided investment allocations. In the early stages of innovation, venture capital is available to fund startups with the promise of outsized return driven partially by technological hype and partially by the opportunity to take large equity stakes in young companies. At the other end of the barbell, private equity and infrastructure capital are available to mature companies seeking an acquisition or project financing based on predictable cash flows and known technologies.
However, gaps appear in the middle as capital requirements increase (often by an order of magnitude) to support the transition to early commercialization. This phenomenon is called the “valley of death” as companies struggle to raise the capital they need to get to scale given the uncertainties they face.

Figure 1. The “valley of death” describes the mismatch between existing financial structures and capital requirements in the crucial early commercialization phase. (Source: Maryland Energy Innovation Accelerator)
Shortcoming of Federal Subsidies
While the federal government has provided loans and subsidies in the past, its programs remain highly reactive and require large amounts of funding.
Aside from asking investors to take on greater risk and lower returns, there are several tools in place to ameliorate the valley of death. The IRA one such example: It appropriated some $370 billion for climate-related spending with a range of instruments, including tax subsidies for renewable energy production, low-cost loans through organizations such as the Department of Energy’s Loan Program Office (LPO), and discretionary grants.
On the other hand, there are major issues with this approach. First, funding is spread out across many calls for funding that tend to be slow, opaque, and costly. Indeed, it is difficult to keep track of available resources, funding announcements, and key requirements—just try searching for a comprehensive, easy-to-understand list of opportunities.
More importantly, these funding mechanisms are simply expensive. The U.S. does not have the financial capacity to support an IRA or CHIPS Act for every industry, nor should it go down that route. While one could argue that these bills reflect the true cost of achieving the stated policy aims of energy transition or securing the semiconductor supply chain, it is also the case that there both knowledge (engineering expertise) and capital (manufacturing facility) capabilities underpin these technologies. Allowing these networks to atrophy created greater costs down the road, which could have been prevented by targeted investments at the right points of development.
The Future Is Dynamic
The future is not perfectly knowable, and new technological needs may arise that change priorities or solve previous problems. Therefore, agility and constant re-evaluation are essential.
Technological progress is not static. Take the concept of peak oil: For decades, many of the world’s most intelligent geologists and energy forecasters believed that the world would quickly run out of oil reserves as the easiest to extract resources were extracted. In reality, technological advances in chemistry, surveying, and drilling enabled hydraulic fracturing (fracking) and horizontal drilling, creating access to “unconventional reserves” that substantially increased fossil fuel supply.

Figure 2. In 1956, M.K. Hubbert created “peak oil” theory, projecting that reserves would be exhausted around the turn of the millennium.
Fracking greatly expanded fossil fuel production in the U.S., increasing resource supply, securing greater energy independence, and facilitating the transition from coal to natural gas, whose expansion has proved to be a helpful bridge towards renewable energy generation. This transition would not have been possible without a series of technological innovations—and highly motivated entrepreneurs—that arose to meet the challenge of energy costs.
To meet the challenges of tomorrow, policymakers need tools that provide them with flexible and targeted options as well as sufficient scale to make an impact on technologies that might need to get through the valley of death. However, they need to remain sufficiently agile so as not to distort well-functioning market forces. This balance is challenging to achieve and requires an organizational structure, authorizations, and funding mechanisms that are sufficiently nimble to adapt to changing technologies and markets.
Opportunity
Given these challenges, it seems unlikely that solutions that rely solely on the private sector will bridge the commercialization gap in a number of capital-intensive strategic industries. On the other hand, existing public-sector tools, such as grants and subsidies, are too costly to implement at scale for every possible externality and are generally too retrospective in nature rather than forward-looking. The government can be an impactful player in bridging the innovation gap, but it needs to do so cost-efficiently.
An SIF is a promising potential solution to the challenges posed above. By its nature, an SIF would have a public mission focused on strategic technologies crossing the valley of death by using targeted interventions and creative financing structures that crowd in private investors. This would enable the government to more sustainably fund innovation, maintain a light touch on private companies, and support key industries and technologies that will define the future global economic and security outlook.
Plan of Action
Recommendation 1. Shepherd technologies through the valley of death.
While the SIF’s investment managers are expected to make the best possible returns, this is secondary to the overarching public policy goal of ensuring that strategically and economically vital technologies have an opportunity to get to commercial scale.
The SIF is meant to crowd in capital such that we achieve broader societal gains—and eventually, market-rate returns—enabled by technologies that would not have survived without timely and well-structured funding. This creates tension between two competing goals: The SIF needs to act as if it will intend to make returns, or else there is the potential for moral hazard and complacency. However, it also has to be willing to not make market-rate returns, or even lose some of its principal, in the service of broader market and ecosystem development.
Thus, it needs to be made explicitly clear from the beginning that an SIF has the intent of achieving market rate returns by catalyzing strategic industries but is not mandated to do so. One way to do this is to adopt a 501(c)(3) structure that has a loose affiliation to a department or agency, similar to that of In-Q-Tel. Excess returns could either be recycled to the fund or distributed to taxpayers.
The SIF should adapt the practices, structures, and procedures of established private-sector funds. It should have a standing investment committee made up of senior stakeholders across various agencies and departments (expanded upon below). Its day-to-day operations should be conducted by professionals who provide a range of experiences, including investing, engineering and technology, and public policy across a spectrum of issue areas.
In addition, the SIF should develop clear underwriting criteria and outputs for each investment. These include, but are not limited to, identifying the broader market and investment thesis, projecting product penetration, and developing potential return scenarios based on different permutations of outcomes. More critically, each investment needs to create a compelling case for why the private sector cannot fund commercialization on its own and why public catalytic funding is essential.
Recommendation 2. The SIF should have a permanent authorization to support innovation under the Department of Commerce.
The SIF should be affiliated with the Department of Commerce but work closely with other departments and agencies, including the Department of Energy, Department of Treasury, Department of Defense, Department of Health and Human Services, National Science Foundation, and National Economic Council.
Strategic technologies do not fall neatly into one sector and cut across many customers. Siloing funding in different departments misses the opportunity to capture funding synergies and, more importantly, develop priorities that are built through information sharing and consensus. Enter the Department of Commerce. In addition to administering the National Institute of Standards and Technology, they have a strong history of working across agencies, such as with the CHIPS Act.
Similar arguments can also be made for the Treasury, and it may even be possible to have Treasury and Commerce work together to manage an SIF. They would be responsible for bringing in subject matter experts (for example, from the Department of Energy or National Science Foundation) to provide specific inputs and arguments for why specific technologies need government-based commercialization funding and at what point such funding is appropriate, acting as an honest broker to allocate strategic capital.
To be clear: The SIF is not intended to supersede any existing funding programs (e.g., the Department of Energy’s Loan Program Office or the National Institute of Health’s ARPA-H) that provide fit-for-purpose funding to specific sectors. Rather, an SIF is intended to fill in the gaps and coordinate with existing programs while providing more creative financing structures than are typically available from government programs.
Recommendation 3. Create a clear innovation roadmap.
Every two years, the SIF should develop or update a roadmap of strategically important industries, working closely with private, nonprofit, and academic experts to define key technological and capability gaps that merit public sector investment.
The SIF’s leaders should be empowered to make decisions on areas to prioritize but have the ability to change and adapt as the economic environment evolves. Although there is a long list of industries that an SIF could potentially support, resources are not infinite. However, a critical mass of investment is required to ensure adequate resourcing. One acute challenge is that this is not perfectly known in advance and changes depending on the technology and sector. However, this is precisely what the strategic investment roadmap is supposed to solve for: It should provide an even-handed assessment of the likely capital requirements and where the SIF is best suited to provide funding compared to other agencies or the private sector.
Moreover, given the ever-changing nature of technology, the SIF should frequently reassess its understanding of key use cases and their broader economic and strategic importance. Thus, after initial development of the SIF, it should be updated every two years to ensure that its takeaways and priorities remain relevant. This is no different than documents such as the National Security Strategy, which are updated every two to four years; in fact, the SIF’s planning documents should flow seamlessly into the National Security Strategy.
To provide a sufficiently broad set of perspectives, the government should include the expertise and insights of outside experts to develop its plan. Existing bodies, such as the President’s Council of Advisors on Science and Technology and the National Quantum Initiative, provide some of the consultative expertise required. However, the SIF should also stand up subject matter specific advisory bodies where a need arises (for example, on critical minerals and mining) and work internally to set specific investment areas and priorities.
Recommendation 4. Limit the SIF to financing.
The government should not be an outsized player in capital markets. As such, the SIF should receive no governance rights (e.g., voting or board seats) in the companies that it invests in.
Although the SIF aims to catalyze technological and ecosystem development, it should be careful not to dictate the future of specific companies. Thus, the SIF should avoid information rights beyond financial reporting. Typical board decks and stockholder updates include updates on customers, technologies, personnel matters, and other highly confidential and specific pieces of information that, if made public through government channels, would play a highly distortionary role in markets. Given that the SIF is primarily focused on supporting innovation through a particularly tricky stage to navigate, the SIF should receive the least amount of information possible to avoid disrupting markets.
Recommendation 5. Focus on providing first-loss capital.
First-loss capital should be the primary mechanism by which the SIF supports new technologies, providing greater incentives for private-sector funders to support early commercialization while providing a means for taxpayers to directly participate in the economic upside of SIF-supported technologies.
Consider the following stylized example to demonstrate a key issue in the valley of death. A promising clean technology company, such as a carbon-free cement or long-duration energy storage firm, is raising $100mm of capital for facility expansion and first commercial deployment. To date, the company has likely raised $30 – $50mm of venture capital to enable tech development, pilot the product, and grow the team’s engineering, R&D, and sales departments.
However, this company faces a fundraising dilemma. Its funding requirements are now too big for all but the largest venture capital firms, who may or may not want to invest in projects and companies like these. On the other hand, this hypothetical company is not mature enough for private equity buyouts nor is it a good candidate for typical project-based debt, which typically require several commercial proof points in order to provide sufficient risk reduction for an investor whose upside is relatively limited. Hence, the “valley of death.”
First-loss capital is an elegant solution to this issue: A prospective funder could commit to equal (pro rata) terms as other investors, except that this first-loss funder is willing to use its investment to make other investors whole (or at least partially offset losses) in the event that the project or company does not succeed. In this example, a first-loss funder would commit to $33.5 million of equity funding (roughly one-third of the company’s capital requirement). If the company succeeds, the first-loss funder makes the same returns as the other investors. However, if the company is unable to fully meet these obligations, the first-loss funder’s $33.5 million would be used to pay the other investors back (the other $66.5 million that was committed). This creates a floor on losses for the non-first-loss investors: Rather than being at risk of losing 100% of their principal, they are at risk of losing 50% of their principal.
The creation of a first-loss layer has a meaningful impact on the risk-reward profile for non-first-loss investors, who now have a floor on returns (in the case above, half their investment). By expanding the acceptable potential loss ratio, growth equity capital (or another appropriate instrument, such as project finance) can fill the rest, thereby crowding in capital.
From a risk-adjusted returns standpoint, this is not a free lunch for the government or taxpayers. Rather, it is intended to be a capital-efficient way of supporting the private-sector ecosystem in developing strategically and economically vital technologies. In other words, it leverages the power of the private sector to solve externalities while providing just enough support to get them to the starting line in the first place.
Conclusion
Many of tomorrow’s strategically important technologies face critical funding challenges in the valley of death. Due to their capital intensity and uncertain outcomes, existing financing tools are largely falling short in the critical early commercialization phases. However, a nimble, properly funded SIF could bridge key gaps while allowing the private sector to do most of the heavy lifting. The SIF would require buy-in from many stakeholders and well-defined sources of funding, but these can be solved with the right mandates, structures, and pay-fors. Indeed, the stakes are too high, and the consequences too dire, to not get strategic innovation right in the 21st century.
This action-ready policy memo is part of Day One 2025 — our effort to bring forward bold policy ideas, grounded in science and evidence, that can tackle the country’s biggest challenges and bring us closer to the prosperous, equitable and safe future that we all hope for whoever takes office in 2025 and beyond.
PLEASE NOTE (February 2025): Since publication several government websites have been taken offline. We apologize for any broken links to once accessible public data.
Put simply, there needs to be an entity that is actually willing and able to absorb lower returns, or even lose some of its principal, in the service of building an ecosystem. Even if the “median” outcome is a market-rate return of capital, the risk-adjusted returns are in effect far lower because the probability of a zero outcome for first-loss providers is substantially nonzero. Moreover, it’s not clear exactly what the right probability estimate should be; therefore, it requires a leap of faith that no economically self-interested private market actor would be willing to take. While some quasi-social-sector organizations can play this role (for example, Bill Gates’s Breakthrough Energy Ventures for climate tech), their capacity is finite, nor is there a guarantee that such a vehicle will appear for every sector of interest. Therefore, a publicly funded SIF is an integral solution to bridging the valley of death.
No, the SIF would not always have to use first-loss structures. However, it is the most differentiated structure that is available to the U.S. government; otherwise, a private-sector player is likely able—and better positioned—to provide funding.
The SIF should be able to use the full range of instruments, including project finance, corporate debt, convertible loans, and equity capital, and all combinations thereof. The instrument of choice should be up to the judgment of the applicant and SIF investment team. This is distinct from providing first-loss capital: Regardless of the financial instrument used, the SIF’s investment would be used to buffer other investors against potential losses.
The target return rate should be commensurate with that of the instrument used. For example, mezzanine debt should target 13–18% IRR, while equity investments should aim for 20–25% IRR. However, because of the increased risk of capital loss to the SIF given its first loss position, the effective blended return should be expected to be lower.
The SIF should be prepared to lose capital on each individual investment and, as a blended portfolio, to have negative returns. While it should underwrite such that it will achieve market-rate returns if successful in crowding in other capital that improves the commercial prospects of technologies and companies in the valley of death, the SIF has a public goal of ecosystem development for strategic domains. Therefore, lower-than-market-rate returns, and even some principal degradation, is acceptable but should be avoided as much as possible through the prudence of the investment committee.
By and large, the necessary public protections are granted through CFIUS, which requires regulatory approval for export or foreign ownership stakes with voting rights above 25% of critical technologies. The SIF can also enact controls around information rights (e.g., customer lists, revenue, product roadmaps) such that they have a veto on parties that can receive such information. However, given its catalytic mission, the SIF does not need board seats or representation and should focus on ensuring that critical technologies and assets are properly protected.
In most private investment firms, the investment committee is made up of the most senior individuals in the fund. These individuals can cross asset classes, sectors of expertise, and even functional backgrounds. However, the investment committee represents a wide breadth of expertise and experiences that, when brought together, enable intellectual honesty and the application of collective wisdom and judgment to the opportunity at hand.
Similarly, the SIF’s investment committee could include the head of the fund and representatives from various departments and agencies in alignment with its strategic priorities. The exact size of the investment committee should be defined by these priorities, but approval should be driven by consensus, and unanimity (or near unanimity) should be expected for investments that are approved.
Given the fluid nature of investment opportunities, the committee should be called upon whenever needed to evaluate a potential opportunity. However, given the generally long process times for investments discussed above (6–12 months), the investment committee should have been briefed multiple times before a formal decision is made.
Check sizes can be flexible to the needs of the investment opportunity. However, as an initial guiding principle, first loss capital should likely make up 20–35% of capital invested so as to require private-sector investors to have meaningful skin in the game. Depending on the fundraise size, this could imply investments of $25 million to $100 million.
Target funding amounts should be set over multiyear timeframes, but the annual appropriations process implies that there will likely be a set cap in any given year. In order to meet the needs of the market, there should be mechanisms that enable emergency draws, up to a cap (e.g., 10% of the annual target funding amount, which will need to be “paid for” by reducing future outlays).
An economically efficient way to fund a government program in support of a positive externality is a Pigouvian tax on negative externalities (such as carbon). However, carbon taxes are as politically unappealing as they are economically sensible and need to be packaged into other policy goals that could potentially support such legislation. Notwithstanding the questionable economic wisdom of tariffs in general, some 56% of voters support a 10% tax on all imports and 60% tariffs on China. Rather than using tariffs harmfully, they could be used more productively. One such proposal is a carbon import tariff that taxes imports on the carbon emitted in the production and transportation of goods into the U.S.
The U.S. would not be a first mover: in fact, the European Union has already implemented a similar mechanism called the Carbon Border Adjustment Mechanism (CBAM), which is focused on heavy industry, including cement, iron and steel, aluminum, fertilizers, electricity, and hydrogen, with chemicals and polymers potentially to be included after 2026. At full rollout in 2030, the CBAM is expected to generate roughly €10–15 billion of tax revenue. Tax receipts of a similar size could be used to fund an SIF or, if Congress authorizes an upfront amount, could be used to nullify the incremental deficit over time.
The EU’s CBAM phased in its reporting requirements over several years. Through July 2024, companies were allowed to use default amounts per unit of production without an explanation as to why actual data was not used. Until January 1, 2026, companies can make estimates for up to 20% of goods; thereafter, the CBAM requires reporting of actual quantities and embedded greenhouse gas emissions.
The U.S. could use a similar phase-in, although given the challenges of carbon reporting, could allow companies to use the lower of actual, verified emissions or per-unit estimates. Under a carbon innovation fee regime, exporters and countries could apply for exemption on a case-by-case basis to the Department of Commerce, which they could approve in line with other goals (e.g., economic development in a region).
The SIF could also be funded by repurposing other funding and elevating their strategic importance. Potential candidates include the Small Business Innovation Research (SBIR) and Small State Business Credit Initiative (SSBCI), which could play a bigger role if moved into the SIF umbrella. For example, the SBIR program, whose latest reporting data is as of FY2019, awarded $3.3 billion in funding that year and $54.6 billion over its lifespan. Moreover, the SSBCI, a $10 billion fund that already provides loan guarantees and other instruments similar to those described above, can be used to support technologies that fall into the purview of the SIF.
Congress could also assess reallocating dollars towards an SIF from spending reforms that are likely inevitable given the country’s fiscal position. In 2023, the Congressional Budget Office (CBO) published a report highlighting potential solutions for reducing the budget deficit. Some potential solutions, like establishing caps on Medicaid federal spending, while fiscally promising, seem unlikely to pass in the near future. However, others are more palatable, especially those that eliminate loopholes or ask higher-income individuals to pay their fair share.
For instance, increasing the amount subject to Social Security taxes above the $250,000 threshold has the potential to raise up to $1.2 trillion over 10 years; while this can be calibrated, an SIF would take only a small fraction of the taxes raised. In addition, the CBO found that federal matching funds for Medicaid frequently ended up getting back to healthcare providers in the form of higher reimbursement rates; eliminating what are effectively kickbacks could reduce the deficit by up to $525 billion over 10 years.
Not Accessible: Federal Policies Unnecessarily Complicate Funding to Support Differently Abled Researchers. We Can Change That.
Persons with disabilities (PWDs) are considered the largest minority in the nation and in the world. There are existing policies and procedures from agencies, directorates, or funding programs that provide support for Accessibility and Accommodations (A&A) in federally funded research efforts. Unfortunately, these policies and procedures all have different requirements, processes, deadlines, and restrictions. This lack of standardization can make it difficult to acquire the necessary support for PWDs by placing the onus on them or their Principal Investigators (PIs) to navigate complex and unique application processes for the same types of support.
This memo proposes the development of a standardized, streamlined, rolling, post-award support mechanism to provide access and accommodations for PWDs as they conduct research and disseminate their work through conferences and convenings. The best case scenario is one wherein a PI or their institution can simply submit the identifying information for the award that has been made and then make a direct request for the support needed for a given PWD to work on the project. In a multi-year award such a request should be possible at any time within the award period.
This could be implemented by a single, streamlined policy adopted by all agencies with the process handled internally. Or, by a new process across agencies under Office of Science and Technology Policy (OSTP) or Office of Management and Budget (OMB) that handles requests for accessibility and accommodations at federally funded research sites and at federally funded convenings. An alternative to a single streamlined policy across these agencies might be a new section in the uniform guidance for federal funding agencies, also known as 2 CFR 200.
This memo focuses on Federal Open Science funding programs to illustrate the challenges in getting A&A funding requests supported. The authors have taken an informal look at agencies outside of science and technology funding. We found similar challenges across federal grantmaking in the Arts and Humanities, Social Services, and Foreign Relations and Aid entities. Similar issues likely exist in private philanthropy as well.
Challenge and Opportunity
Deaf/hard-of-hearing (DHH), Blind/low-vision (BLV), and other differently abled academicians, senior personnel, students, and post-doctoral fellows engaged in federally funded research face challenges in acquiring accommodations for accessibility. These include, but are not limited to:
- Human-provided ASL-English interpreting and interview transcription services for the DHH and non-DHH participants. While there are some applications of artificial intelligence (AI) that show promise on the transcription side, there’s a long way to go on ASL interpretation in an AI provided model versus the use of human interpreters.
- Visual and Pro-tactile interpreting/descriptive services for the BLV participants
- Adaptive lab equipment and computing peripherals
- Accessibility support or remediation for physical sites
Having these services available is crucial for promoting an inclusive research environment on a larger scale.
Moving to a common, post-award process:
- Allows the PI and the reviewers more time and space to focus on the core research efforts being described in the initial proposal
- Removes any chance of the proposal funding being taken out of consideration due to higher costs in comparison to similar proposals in the pool
- Creates a standard, replicable pathway for seeking accommodations once the overall proposal has been funded. This is especially true if the support comes from a single process across all federal funding programs rather than within each agency.
- Allows for flexibility in accommodations. Needs vary from person-to-person and case-to-case. For example, in the case of workplace accommodations for DHH team members, one full-time researcher may request full-time ASL interpretation on-site, while another might prefer to work primarily through digital text channels; only requiring ASL interpretation for staff meetings and other group activities.
- Potentially reduces federal government financial and human resources currently expended in supporting such requests by eliminating duplication of effort across agencies or, at minimum streamlining processes within agencies.
Such a process might follow these steps below. The example below is from the National Science Foundation (NSF), but the same, or similar process could be done within any agency:
- PI receives notification of grant award from NSF. PI identifies need for A & A services at start, or at any time during the grant period
- PI (or SRS staff) submits request for A&A funding support to NSF. Request includes NSF program name and award number, the specifics of the requested A & A support, a budget justification and three vendor quotes (if needed)
- Use of funds is authorized, and funding is released to PI’s institution and acquisition would follow their standard purchasing or contracting procedures
- PI submits receipts/ paid vendor invoice to funding body
- PI cites and documents use of funds in annual report, or equivalent, to NSF
Current Policies and Practices
Pre-Award Funding
Principal Investigators (PIs) who request A&A support for themselves or for other members of the research team are sometimes required to apply for it in their initial grant proposals. This approach has several flaws.
First and foremost, this funding process reduces the direct application of research dollars for these PIs and their teams compared to other researchers in the same program. Simply put, if two applicants are applying for a $100,000 grant, and one needs to fund $10,000 worth of accommodations, services, and equipment out of the award, they have $10,000 less to pursue the proposed research activities. This essentially creates a “10% A & A tax” on the overall research funding request.
Lived Experience Example
In a real world example, the author and his colleague, the late Dr. Mel Chua, were awarded a $60,000, one year grant to do a qualitative research case study as part of the Ford Foundation Critical Digital Infrastructure Research cohort. As Dr. Chua was Deaf, the PIs pointed out to Ford that $10,000 worth of support services would be needed to cover costs for
- American Sign Language (ASL) interpreters during the qualitative interviews and advisory committee meetings
- Transcription of the interviews
- ASL Interpreting for conference dissemination and collection of comments at formal and informal meetings during those conferences
We communicated the fact that spending general research award money on those services would reduce the research work the funds were awarded to support. The Ford Foundation understood and provided an additional $10,000 as post-award funding to cover those services. Ford did not inform the PIs as to whether that support came from another directed set of funds for A&A support or from discretionary dollars within the foundation.
Second, it can be limiting for the funded project to work with or hire PWDs as co-PIs, students, or if they weren’t already part of the original grant proposal. For example, suppose a research project is initially awarded funding for four years without A&A support and then a promising team member who is a PWD appears on the scene in year three who would require it. In this case, PIs then must:
- Reallocate research dollars meant for other uses within the grant to support A&A;
- Find other funding to support those needs within their institution;
- Navigate the varied post-award support landscape, sometimes going so far as to write an entirely new full proposal with a significant review timeline, to try to get support. If this happens off cycle, the funding might not even arrive until the last few months of the fourth year.
- Or, not hire the person in question because they can’t provide the needed A&A.
Post-Award Funding
Some agencies have programs for post-award supplemental funding that address the challenges described above. While these are well-intentioned, many are complicated and often have different timelines, requirements, etc. In some cases, a single supplemental funding source may be addressing all aspects of diversity, equity and inclusion as well as A&A. The needs and costs in the first three categories are significantly different than in the last. Some post-award pools come from the same agency’s annual allocation program-wide. If those funds have been primarily expended on the initial awards for the solicitation, there may be little, or no money left to support post-award funding for needed accommodations. The table below briefly illustrates the range of variability across a subset of representative supplemental funding programs. There are links in the top row of the table to access the complete program information. Beyond the programs in this table, more extensive lists of NSF and NIH offerings are provided by those agencies. One example is the NSF Dear Colleague Letter Persons with Disabilities – STEM Engagement and Access.
Ideally these policies and procedures, and others like them, would be replaced by a common, post-award process. PIs or their institutions would simply submit the identifying information on the grant that had been awarded and the needs for Accommodations and Accessibility to support team members with disabilities at any time during the grant period.
Plan of Action
The OSTP, possibly in a National Science and Technology Council interworking group process,, should conduct an internal review of the A&A policies and procedures for grant programs from federal scientific research aligned agencies. This could be led by OSTP directly or under their auspices and led by either NSF or the National Institute of Health (NIH). Participants would be relevant personnel from DOE, DOD, NASA, USDA, EPA, NOAA, NIST and HHS, at minimum. The goal should be to create a draft of a single, streamlined policy and process, post-award, for all federal grant programs or a new section in the uniform guidance for federal funding agencies.
There should be an analysis of the percentages, size and amounts of awards currently being made to support A&A in research funding grant programs. It’s not clear how the various funding ranges and caps listed in the table above were determined or if they meet the needs. One goal of this analysis would be to determine how well current needs within and across agencies are being met and what future needs might be.
A second goal would be to look at the level of duplication of effort and scope of manpower savings that might be attained by moving to a single, streamlined policy. This might be a coordinated process between OMB and OSTP or a separate one done by OMB. No matter how it is coordinated, an understanding of these issues should inform whatever new policies or new additions to 2 CFR 200 would emerge.
A third goal of this evaluation could be to consider if the support for A&A post-award funding might best be served by a single entity across all federal grants, consolidating the personnel expertise and policy and process recommendations in one place. It would be a significant change, and could require an act of Congress to achieve, but from the point of view of the authors it might be the most efficient way to serve grantees who are PWDs.
Once the initial reviews as described above, or a similar process is completed, the next step should be a convening of stakeholders outside of the federal government with the purpose of providing input to the streamlined draft policy. These stakeholder entities could include, but should not be limited to, the National Association for the Deaf, The American Foundation for the Blind, The American Association of People with Disabilities and the American Diabetes Association. One of the goals of that convening should be a discussion, and decision, as to whether a period of public comment should be put in place as well, before the new policy is adopted.
Conclusion
The above plan of action should be pursued so that more PWDS will be able to participate, or have their participation improved, in federally funded research. A policy like the one described above lays the groundwork and provides a more level playing field for Open Science to become more accessible and accommodating.It also opens the door for streamlined processes, reduced duplication of effort and greater efficiency within the engine of Federal Science support.
Acknowledgments
The roots of this effort began when the author and Dr. Mel Chua and Stephen Jacobs received funding for their research as part of the first Critical Digital Infrastructure research cohort and were able to negotiate for accessibility support services outside their award. Those who provided input on the position paper this was based on are:
- Dr. Mel Chua, Independent Researcher
- Dr. Liz Hare, Quantitative Geneticist, Dog Genetics LLC
- Dr. Christopher Kurz, Professor and Director of Mathematics and Science Language and Learning Lab, National Technical Institute for the Deaf
- Luticha Andre-Doucette, Catalyst Consulting
This action-ready policy memo is part of Day One 2025 — our effort to bring forward bold policy ideas, grounded in science and evidence, that can tackle the country’s biggest challenges and bring us closer to the prosperous, equitable and safe future that we all hope for whoever takes office in 2025 and beyond.
PLEASE NOTE (February 2025): Since publication several government websites have been taken offline. We apologize for any broken links to once accessible public data.
Based on the percentage of PWDs in the general population size, conference funders should assume that some of their presenters or attendees will need accommodations. Funding from federal agencies should be made available to provide an initial minimum-level of support for necessary A & A. The event organizers should be able to apply for additional support above the minimum level if needed, provided participant requests are made within a stated time before the event. For example, a stipulated deadline of six weeks before the event to request supplemental accommodation, so that the organizers can acquire what’s needed within thirty days of the event.
Yes, in several ways. In general, most of the support needed for these is in service provision vs. hardware/software procurement. However, understanding the breadth and depth of issues surrounding human services support is more complex and outside the experience of most PIs running a conference in their own scientific discipline.
Again, using the example of DHH researchers who are attending a conference. A conference might default to providing a team of two interpreters during the conference sessions, as two per hour is the standard used. Should a group of DHH researchers attend the conference and wish to go to different sessions or meetings during the same convening, the organizers may not have provided enough interpreters to support those opportunities.
By providing interpretation for formal sessions only, DHH attendees are excluded from a key piece of these events, conversations outside of scheduled sessions. This applies to both formally planned and spontaneous ones. They might occur before, during, or after official sessions, during a meal offsite, etc. Ideally interpreters would be provided for these as well.
These issues, and others related to other groups of PWDs, are beyond the experience of most PIs who have received event funding.
There are some federal agency guides produced for addressing interpreting and other concerns, such as the “Guide to Developing a Language Access Plan” Center for Medicare and Medicaid Services (CMS). These are often written to address meeting needs of full-time employees on site in office settings. These generally cover various cases not needed by a conference convener and may not address what they need for their specific use case. It might be that the average conference chair and their logistics committee is a simply stated set of guidelines to address their short-term needs for their event. Additionally, a directory of where to hire providers with the appropriate skill sets and domain knowledge to meet the needs of PWDs attending their events would be an incredible aid to all concerned.
The policy review process outlined above should include research to determine a base level of A & A support for conferences. They might recommend a preferred federal guide to these resources or identify an existing one.
Promoting Fairness in Medical Innovation
There is a crisis within healthcare technology research and development, wherein certain groups due to their age, gender, or race and ethnicity are under-researched in preclinical studies, under-represented in clinical trials, misunderstood by clinical practitioners, and harmed by biased medical technology. These issues in turn contribute to costly disparities in healthcare outcomes, leading to losses of $93 billion a year in excess medical-care costs, $42 billion a year in lost productivity, and $175 billion a year due to premature deaths. With the rise of artificial intelligence (AI) in healthcare, there’s a risk of encoding and recreating existing biases at scale.
The next Administration and Congress must act to address bias in medical technology at the development, testing and regulation, and market-deployment and evaluation phases. This will require coordinated effort across multiple agencies. In the development phase, science funding agencies should enforce mandatory subgroup analysis for diverse populations, expand funding for under-resourced research areas, and deploy targeted market-shaping mechanisms to incentivize fair technology. In the testing and regulation phase, the FDA should raise the threshold for evaluation of medical technologies and algorithms and expand data-auditing processes. In the market-deployment and evaluation phases, infrastructure should be developed to perform impact assessments of deployed technologies and government procurement should incentivize technologies that improve health outcomes.
Challenge and Opportunity
Bias is regrettably endemic in medical innovation. Drugs are incorrectly dosed to people assigned female at birth due to historical exclusion of women from clinical trials. Medical algorithms make healthcare decisions based on biased health data, clinically disputed race-based corrections, and/or model choices that exacerbate healthcare disparities. Much medical equipment is not accessible, thus violating the Americans with Disabilities Act. And drugs, devices, and algorithms are not designed with the lifespan in mind, impacting both children and the elderly. Biased studies, technology, and equipment inevitably produce disparate outcomes in U.S. healthcare.
The problem of bias in medical innovation manifests in multiple ways: cutting across technological sectors in clinical trials, pervading the commercialization pipeline, and impeding equitable access to critical healthcare advances.
Bias in medical innovation starts with clinical research and trials
The 1993 National Institutes of Health (NIH) Revitalization Act required federally funded clinical studies to (i) include women and racial minorities as participants, and (ii) break down results by sex and race or ethnicity. As of 2019, the NIH also requires inclusion of participants across the lifespan, including children and older adults. Yet a 2019 study found that only 13.4% of NIH-funded trials performed the mandatory subgroup analysis, and challenges in meeting diversity targets continue into 2024 . Moreover, the increasing share of industry-funded studies are not subject to Revitalization Act mandates for subgroup analysis. These studies frequently fail to report differences in outcomes by patient population as a result. New requirements for Diversity Action Plans (DAPs), mandated under the 2023 Food and Drug Omnibus Reform Act, will ensure drug and device sponsors think about enrollment of diverse populations in clinical trials. Yet, the FDA can still approve drugs and devices that are not in compliance with their proposed DAPs, raising questions around weak enforcement.
The resulting disparities in clinical-trial representation are stark: African Americans represent 12% of the U.S. population but only 5% of clinical-trial participants, Hispanics make up 16% of the population but only 1% of clinical trial participants, and sex distribution in some trials is 67% male. Finally, many medical technologies approved prior to 1993 have not been reassessed for potential bias. One outcome of such inequitable representation is evident in drug dosing protocols: sex-aware prescribing guidelines exist for only a third of all drugs.
Bias in medical innovation is further perpetuated by weak regulation
Algorithms
Regulation of medical algorithms varies based on end application, as defined in the 21st Century Cures Act. Only algorithms that (i) acquire and analyze medical data and (ii) could have adverse outcomes are subject to FDA regulation. Thus, clinical decision-support software (CDS) is not regulated even though these technologies make important clinical decisions in 90% of U.S. hospitals. The FDA has taken steps to try and clarify what CDS must be considered a medical device, although these actions have been heavily criticized by industry. Finally, the lack of regulatory frameworks for generative AI tools is leading to proliferation without oversight.
Even when a medical algorithm is regulated, regulation may occur through relatively permissive de novo pathways and 510(k) pathways. A de novo pathway is used for novel devices determined to be low to moderate risk, and thus subject to a lower burden of proof with respect to safety and equity. A 510(k) pathway can be used to approve a medical device exhibiting “substantial equivalence” to a previously approved device, i.e., it has the same intended use and/or same technological features. Different technical features can be approved so long as there are no questions raised around safety and effectiveness.
Medical algorithms approved through de novo pathways can be used as predicates for approval of devices through 510(k) pathways. Moreover, a device approved through a 510(k) pathway can remain on the market even if its predicate device was recalled. Widespread use of 510(k) approval pathways has generated a “collapsing building” phenomenon, wherein many technologies currently in use are based on failed predecessors. Indeed, 97% of devices recalled between 2008 to 2017 were approved via 510(k) clearance.
While DAP implementation will likely improve these numbers, for the 692 AI-ML enabled medical devices, only 3.6% reported race or ethnicity, 18.4% reported age, and only .9% include any socioeconomic information. Further, less than half did detailed analysis of algorithmic performance and only 9% included information on post-market studies, raising the risk of algorithmic bias following approvals and broad commercialization.
Even more alarming is evidence showing that machine learning can further entrench medical inequities. Because machine learning medical algorithms are powered by data from past medical decision-making, which is rife with human error, these algorithms can perpetuate racial, gender, and economic bias. Even algorithms demonstrated to be ‘unbiased’ at the time of approval can evolve in biased ways over time, with little to no oversight from the FDA. As technological innovation progresses, especially generative AI tools, an intentional focus on this problem will be required.
Medical devices
Currently, the Medical Device User Fee Act requires the FDA to consider the least burdensome appropriate means for manufacturers to demonstrate the effectiveness of a medical device or to demonstrate a device’s substantial equivalence. This requirement was reinforced by the 21st Century Cures Act, which also designated a category for “breakthrough devices” subject to far less-stringent data requirements. Such legislation shifts the burden of clinical data collection to physicians and researchers, who might discover bias years after FDA approval. This legislation also makes it difficult to require assessments on the differential impacts of technology.
Like medical algorithms, many medical devices are approved through 510(k) exemptions or de novo pathways. The FDA has taken steps since 2018 to increase requirements for 510(k) approval and ensure that Class III (high-risk) medical devices are subject to rigorous pre-market approval, but problems posed by equivalence and limited diversity requirements remain.
Finally, while DAPs will be required for many devices seeking FDA approval, the recommended number of patients in device testing is shockingly low. For example, currently, only 10 people are required in a study of any new pulse oximeter’s efficacy and only 2 of those people need to be “darkly pigmented”. This requirement (i) does not have the statistical power necessary to detect differences between demographic groups, and (i) does not represent the composition of the U.S. population. The standard is currently under revision after immense external pressure. FDA-wide, there are no recommended guidelines for addressing human differences in device design, such as pigmentation, body size, age, and pre-existing conditions.
Pharmaceuticals
The 1993 Revitalization Act strictly governs clinical trials for pharmaceuticals and does not make recommendations for adequate sex or genetic diversity in preclinical research. The results are that a disproportionately high number of male animals are used in research and that only 5% of cell lines used for pharmaceutical research are of African descent. Programs like All of Us, an effort to build diverse health databases through data collection, are promising steps towards improving equity and representation in pharmaceutical research and development (R&D). But stronger enforcement is needed to ensure that preclinical data (which informs function in clinical trials) reflects the diversity of our nation.
Bias in medical innovation are not tracked post-regulatory approval
FDA-regulated medical technologies appear trustworthy to clinicians, where the approval signals safety and effectiveness. So, when errors or biases occur (if they are even noticed), the practitioner may blame the patient for their lifestyle rather than the technology used for assessment. This in turn leads to worse clinical outcomes as a result of the care received.
Bias in pulse oximetry is the perfect case study of a well-trusted technology leading to significant patient harm. During the COVID-19 pandemic, many clinicians and patients were using oximeter technology for the first time and were not trained to spot factors, like melanin in the skin, that cause inaccurate measurements and impact patient care. Issues were largely not attributed to the device. This then leads to underreporting of adverse events to the FDA — which is already a problem due to the voluntary nature of adverse-event reporting.
Even when problems are ultimately identified, the federal government is slow to respond. The pulse oximeter’s limitations in monitoring oxygenation levels across diverse skin tones was identified as early as the 1990s. 34 years later, despite repeated follow-up studies indicating biases, no manufacturer has incorporated skin-tone-adjusted calibration algorithms into pulse oximeters. It required the large Sjoding study, and the media coverage it garnered around delayed care and unnecessary deaths, for the FDA to issue a safety communication and begin reviewing the regulation.
Other areas of HHS are stepping up to address issues of bias in deployed technologies. A new ruling by the HHS Office of Civil Rights (OCR) on Section 1557 of the Affordable Care Act requires covered providers and institutions (i.e. any receiving federal funding) to identify their use of patient care decision support tools that directly measure race, color, national origin, sex, age, or disability, and to make reasonable efforts to mitigate the risk of discrimination from their use of these tools. Implementation of this rule will depend on OCR’s enforcement, and yet it provides another route to address bias in algorithmic tools.
Differential access to medical innovation is a form of bias
Americans face wildly different levels of access to new medical innovations. As many new innovations have high cost points, these drugs, devices, and algorithms exist outside the price range of many patients, smaller healthcare institutions and federally funded healthcare service providers, including the Veterans Health Administration, federally qualified health centers and the Indian Health Service. Emerging care-delivery strategies might not be covered by Medicare and Medicaid, meaning that patients insured by CMS cannot access the most cutting-edge treatments. Finally, the shift to digital health, spurred by COVID-19, has compromised access to healthcare in rural communities without reliable broadband access.
Finally, the Advanced Research Projects Agency for Health (ARPA-H) has a commitment to have all programs and projects consider equity in their design. To fulfill ARPA-H’s commitment, there is a need for action to ensure that medical technologies are developed fairly, tested with rigor, deployed safely, and made affordable and accessible to everyone.
Plan of Action
The next Administration should launch “Healthcare Innovation for All Americans” (HIAA), a whole of government initiative to improve health outcomes by ensuring Americans have access to bias-free medical technologies. Through a comprehensive approach that addresses bias in all medical technology sectors, at all stages of the commercialization pipeline, and in all geographies, the initiative will strive to ensure the medical-innovation ecosystem works for all. HIAA should be a joint mandate of Health and Human Services (HHS) and the Office of Science Technology and Policy (OSTP) to work with federal agencies on priorities of equity, non-discrimination per Section 1557 of the Affordable Care Act and increasing access to medical innovation, and initiative leadership should sit at both HHS and OSTP.
This initiative will require involvement of multiple federal agencies, as summarized in the table below. Additional detail is provided in the subsequent sections describing how the federal government can mitigate bias in the development phase; testing, regulation, and approval phases; and market deployment and evaluation phases.
Three guiding principles should underlie the initiative:
- Equity and non-discrimination should drive action. Actions should seek to improve the health of those who have been historically excluded from medical research and development. We should design standards that repair past exclusion and prevent future exclusion.
- Coordination and cooperation are necessary. The executive and legislative branches must collaborate to address the full scope of the problem of bias in medical technology, from federal processes to new regulations. Legislative leadership should task the Government Accountability Office (GAO) to engage in ongoing assessment of progress towards the goal of achieving bias-free and fair medical innovation.
- Transparent, evidence-based decision making is paramount. There is abundant peer-reviewed literature that examines bias in drugs, devices, and algorithms used in healthcare settings — this literature should form the basis of a non-discrimination approach to medical innovation. Gaps in evidence should be focused on through deployed research funding. Moreover, as algorithms become ubiquitous in medicine, every effort should be made to ensure that these algorithms are trained on representative data of those experiencing a given healthcare condition.
Addressing bias at the development phase
The following actions should be taken to address bias in medical technology at the innovation phase:
- Enforce parity in government-funded research. For clinical research, NIH should examine the widespread lack of adherence to regulations requiring that government funded clinical trials report sex, racial or ethnicity, and age breakdown of trial participants. Funding should be reevaluated for non-compliant trials. For preclinical research, NIH should require gender parity in animal models and representation of diverse cell lines used in federally funded studies.
- Deploy funding to address research gaps. Where data sources for historically marginalized people are lacking, such as for women’s cardiovascular health, NIH should deploy strategic, targeted funding programs to fill these knowledge gaps. This could build on efforts like the Initiative on Women’s Health Research. Increased funding should include resources for underrepresented groups to participate in research and clinical trials through building capacity in community organizations. Results should be added to a publicly available database so they can be accessed by designers of new technologies. Funding programs should also be created to fill gaps in technology, such as in diagnostics and treatments for high-prevalence and high-burden uterine diseases like endometriosis (found in 10% of reproductive-aged people with uteruses).
- Invest in research into healthcare algorithms and databases. Given the explosion of algorithms in healthcare decision-making, NIH and NSF should launch a new research program focused on the study, evaluation, and application of algorithms in healthcare delivery, and on how artificial intelligence and machine learning (AI/ML) can exacerbate healthcare inequities. The initial request for proposals should focus on design strategies for medical algorithms that mitigate bias from data or model choices.
- Task ARPA-H with developing metrics for equitable medical technology development. ARPA-H should prioritize developing a set of procedures and metrics for equitable development of medical technology. Once developed, these processes should be rapidly deployed across ARPA-H, as well as published for potential adoption by additional federal agencies, industry, and other stakeholders. ARPA-H could also collaborate with NIST on standards setting with NIST and ASTP on relevant standards setting. For instance, NIST has developed an AI Risk Management Framework and the ONC engages in setting standards that achieve equity by design. CMS could use resultant standards for Medicare and Medicaid reimbursements.
- Leverage procurement as a demand-signal for medical technologies that work for diverse populations. As the nation’s largest healthcare system, the Veterans Health Administration (VHA) can generate demand-signals for bias-free medical technologies through its procurement processes and market-shaping mechanisms. For example, the VA could put out a call for a pulse oximeter that works equally well across the entire range of human skin pigmentation and offer contracts for the winning technology.
Addressing bias at the testing, regulation, and approval phases
The following actions should be taken to address bias in medical innovation at the testing, regulation, and approval phases:
- Raise the threshold for FDA evaluation of devices and algorithms. Equivalency necessary to receive 510(k) clearance should be narrowed. For algorithms, this would involve consideration of whether the datasets for machine learning tactics used by the new device and its predicate are similar. For devices (including those that use algorithms), this would require tightening the definition of “same intended use” (currently defined as a technology having the same functionality as one previously approved by the FDA) as well as eliminating the approval of new devices with “different technological characteristics” (the application of one technology to a new area of treatment in which that technology is untested).
- Evaluate FDA’s guidance on specific technology groups for equity. Requirements for the safety of a given drug, medical device, or algorithm should have the statistical power necessary to detect differences between demographic groups and represent all end-users of the technology..
- Establish a data bank for auditing medical algorithms. The newly established Office of Digital Transformation within the FDA should create a “data bank” of healthcare images and datasets representative of the U.S. population, which could be done in partnership with the All of Us program. Medical technology developers could use the data bank to assess the performance of medical algorithms across patient populations. Regulators could use the data bank to ground claims made by those submitting a technology for FDA approval.
- Allow data submitted to the FDA to be examined by the broader scientific community. Currently, data submitted to the FDA as part of its regulatory-approval process is kept as a trade secret and not released pre-authorization to researchers. Releasing the data via an FDA-invited “peer review” step in the regulation of high-risk technologies, like automated decision-making algorithms, Class III medical devices, and drugs, will ensure that additional, external rigor is applied to the technologies that could cause the most harm due to potential biases.
- Establish an enforceable AI Bill of Rights. The federal government and Congress should create protections for necessary uses of artificial intelligence (AI) identified by OSTP. Federally funded healthcare centers, like facilities part of the Veterans Health Administration, could refuse to buy software or technology products that violate this “AI Bill of Rights” through changes to federal acquisition regulation (FAR).
Addressing bias at the market deployment and evaluation phases
- Strengthen reporting mechanisms at the FDA. Healthcare providers, who are often closest to the deployment of medical technologies, should be made mandatory reporters to the FDA of all witnessed adverse events related to bias in medical technology. In addition, the FDA should require the inclusion of unique device identifiers (UDIs) in adverse-response reporting. Using this data, Congress should create a national and publicly accessible registry that uses UDIs to track post-market medical outcomes and safety.
- Require impact assessments of deployed technologies. Congress must establish systems of accountability for medical technologies, like algorithms, that can evolve over time. Such work could be done by passing the Algorithmic Accountability Act which would require companies that create “high-risk automated decision systems” to conduct impact assessments reviewed by the FTC as frequently as necessary.
- Assess disparities in patient outcomes to direct technical auditing. AHRQ should be given the funding needed to fully investigate patient-outcome disparities that could be caused by biases in medical technology, such as its investigation into the impacts of healthcare algorithms on racial and ethnic disparities. The results of this research should be used to identify technologies that the FDA should audit post-market for efficacy or the FTC should investigate. CMS and its accrediting agencies can monitor these technologies and assess whether they should receive Medicare and Medicaid funding.
- Review reimbursement guidelines that are dependent on medical technologies with known bias. CMS should review its national coverage determinations for technologies, like pulse oximetry, that are known to perform differently across populations. For example, pulse oximeters can be used to determine home oxygen therapy provision, thus potentially excluding darkly-pigmented populations from receiving this benefit.
- Train physicians to identify bias in medical technologies and identify new areas of specialization. ED should work with medical schools to develop curricula training physicians to identify potential sources of bias in medical technologies and ensuring that physicians understand how to report adverse events to the FDA. In addition, ED should consider working with the American Medical Association to create new medical specialties that work at the intersection of technology and care delivery.
- Ensure that technologies developed by ARPA-H have an enforceable access plan. ARPA-H will produce cutting-edge technologies that must be made accessible to all Americans. ARPA-H should collaborate with the Center for Medicare and Medicaid Innovation to develop strategies for equitable delivery of these new technologies. A cost-effective deployment strategy must be identified to service federally-funded healthcare institutions like Veterans Health Administration hospitals and clinical, federally qualified health centers, and Indian Health Service.
- Create a fund to support digital health technology infrastructure in rural hospitals. To capitalize on the $65 billion expansion of broadband access allocated in the Bipartisan Infrastructure Bill, HRSA should deploy strategic funding to federally qualified health centers and rural health clinics to support digital health strategies — such as telehealth and mobile health monitoring — and patient education for technology adoption.
A comprehensive road map is needed
The GAO should conduct a comprehensive investigation of “black box” medical technologies utilizing algorithms that are not transparent to end users, medical providers, and patients. The investigation should inform a national strategic plan for equity and non-discrimination in medical innovation that relies heavily on algorithmic decision-making. The plan should include identification of noteworthy medical technologies leading to differential healthcare outcomes, creation of enforceable regulatory standards, development of new sources of research funding to address knowledge gaps, development of enforcement mechanisms for bias reporting, and ongoing assessment of equity goals.
Timeline for action
Realizing HIAA will require mobilization of federal funding, introduction of regulation and legislation, and coordination of stakeholders from federal agencies, industry, healthcare providers, and researchers around a common goal of mitigating bias in medical technology. Such an initiative will be a multi-year undertaking and require funding to enact R&D expenditures, expand data capacity, assess enforcement impacts, create educational materials, and deploy personnel to staff all the above.
Near-term steps that can be taken to launch HIAA include issuing a public request for information, gathering stakeholders, engaging the public and relevant communities in conversation, and preparing a report outlining the roadmap to accomplishing the policies outlined in this memo.
Conclusion
Medical innovation is central to the delivery of high-quality healthcare in the United States. Ensuring equitable healthcare for all Americans requires ensuring that medical innovation is equitable across all sectors, phases, and geographies. Through a bold and comprehensive initiative, the next Administration can ensure that our nation continues leading the world in medical innovation while crafting a future where healthcare delivery works for all.
This action-ready policy memo is part of Day One 2025 — our effort to bring forward bold policy ideas, grounded in science and evidence, that can tackle the country’s biggest challenges and bring us closer to the prosperous, equitable and safe future that we all hope for whoever takes office in 2025 and beyond.
PLEASE NOTE (February 2025): Since publication several government websites have been taken offline. We apologize for any broken links to once accessible public data.
HIAA will be successful when medical policies, projects, and technologies yield equitable health care access, treatment, and outcomes. For instance, success would yield the following outcomes:
- Representation in preclinical and clinical research equivalent to the incidence of a studied condition in the general population.
- Research on a disease condition funded equally per affected patient.
- Existence of data for all populations facing a given disease condition.
- Medical algorithms that have equal efficacy across subgroup populations.
- Technologies that work equally well in testing as they do when deployed to the market.
- Healthcare technologies made available and affordable to all care facilities.
Regulation alone cannot close the disparity gap. There are notable gaps in preclinical and clinical research data for women, people of color, and other historically underrepresented groups that need to be filled. There are also historical biases encoded in AI/ML decision making algorithms that need to be studied and rectified. In addition, the FDA’s role is to serve as a safety check on new technologies — the agency has limited oversight over technologies once they are out on the market due to the voluntary nature of adverse reporting mechanisms. This means that agencies like the FTC and CMS need to be mobilized to audit high-risk technologies once they reach the market. Eliminating bias in medical technology is only possible through coordination and cooperation of federal agencies with each other as well as with partners in the medical device industry, the pharmaceutical industry, academic research, and medical care delivery.
A significant focus of the medical device and pharmaceutical industries is reducing the time to market for new medical devices and drugs. Imposing additional requirements for subgroup analysis and equitable use as part of the approval process could work against this objective. On the other hand, ensuring equitable use during the development and approval stages of commercialization will ultimately be less costly than dealing with a future recall or a loss of Medicare or Medicaid eligibility if discriminatory outcomes are discovered.
Healthcare disparities exist in every state in America and are costing billions a year in economic growth. Some of the most vulnerable people live in rural areas, where they are less likely to receive high-quality care because costs of new medical technologies are too high for the federally qualified health centers that serve one in five rural residents as well as rural hospitals. Furthermore, during continued use, a biased device creates adverse healthcare outcomes that cost taxpayers money. A technology functioning poorly due to bias can be expensive to replace. It is economically imperative to ensure technology works as expected, as it leads to more effective healthcare and thus healthier people.
U.S. Energy Security Compacts: Enhancing American Leadership and Influence with Global Energy Investment
This policy proposal was incubated at the Energy for Growth Hub and workshopped at FAS in May 2024.
Increasingly, U.S. national security priorities depend heavily on bolstering the energy security of key allies, including developing and emerging economies. But U.S. capacity to deliver this investment is hamstrung by critical gaps in approach, capability, and tools.
The new administration should work with Congress to give the Millennium Challenge Corporation (MCC) the mandate and capacity to lead the U.S. interagency in implementing ‘Energy Security Compacts’, bilateral packages of investment and support for allies whose energy security is closely tied to core U.S. priorities. This would require minor amendments to the Millennium Challenge Act of 2003 to add a fourth business line to MCC’s Compact operations and grant the agency authority to coordinate an interagency working group contributing complementary tools and resources.
This proposal presents an opportunity to deliver on global energy security, an issue with broad appeal and major national security benefits. This initiative would strengthen economic partnerships with allies overseas, who consistently rank energy security as a top priority; enhance U.S. influence and credibility in advancing global infrastructure; and expand growing markets for U.S. energy technology. This proposal is built on the foundations and successes of MCC, a signature achievement of the G.W. Bush administration, and is informed by lessons learned from other initiatives launched by previous presidents of both parties.
Challenge and Opportunity
More than ever before, U.S. national security depends on bolstering the energy security of key allies. Core examples include:
- Securing physical energy assets: In countries under immediate or potential military threat, the U.S. may seek to secure vulnerable critical energy infrastructure, restore energy services to local populations, and build a foundation for long-term restoration.
- Countering dependence on geostrategic competitors: U.S. allies’ reliance on geostrategic competitors for energy supply or technologies poses short- and long-term threats to national security. Russia is building large nuclear reactors in major economies including Turkey, Egypt, India, and Bangladesh; has signed agreements to supply nuclear technology to at least 40 countries; and has agreed to provide training and technical assistance to at least another 14. Targeted U.S. support, investment, and commercial diplomacy can head off such dependence by expanding competition.
- Driving economic growth and enduring diplomatic relationships: Many developing and emerging economies face severe challenges in providing reliable, affordable electricity to their populations. This hampers basic livelihoods; constrains economic activity, job creation, and internet access; and contributes to deteriorating economic conditions driving instability and unrest. Of all the constraints analyses conducted by MCC since its creation, roughly half identified energy as a country’s top economic constraint. As emerging economies grow, their economic stability has an expanding influence over global economic performance and security. In coming decades, they will require vast increases in reliable energy to grow their manufacturing and service industries and employ rapidly growing populations. U.S. investment can provide the foundation for market-driven growth and enduring diplomatic partnerships.
- Diversifying supply chains: Many crucial technologies depend on minerals sourced from developing economies without reliable electricity. For example, Zambia accounts for about 4% of global copper supply and would like to scale up production. But recurring droughts have shuttered the country’s major hydropower plant and led to electricity outages, making it difficult for mining operations to continue or grow. Scaling up the mining and processing of key minerals in developing economies will require investment in improving power supply.
The U.S. needs a mechanism that enables quick, efficient, and effective investment and policy responses to the specific concerns facing key allies. Currently, U.S. capacity to deliver such support is hamstrung by key gaps in approach, capabilities, and tools. The most salient challenges include:
A project-by-project approach limits systemic impact: U.S. overseas investment agencies including the Development Finance Corporation (DFC), the U.S. Trade and Development Agency (USTDA), and the Export-Import Bank (EXIM) are built to advance individual commercial energy transactions across many different countries. This approach has value–but is insufficient in cases where the goal is to secure a particular country’s entire energy system by building strong, competitive markets. That will require approaching the energy sector as a complex and interconnected system, rather than a set of stand-alone transactions.
Diffusion of tools across the interagency hinders coordination. The U.S. has powerful tools to support energy security–including through direct investment, policy support, and technical and commercial assistance–but they are spread across at least nine different agencies. Optimizing deployment will require efficient coordination, incentives for collaboration; and less fragmented engagement with private partners.
Insufficient leverage to incentivize reforms weakens accountability. Ultimately, energy security depends heavily on decisions made by the partner country’s government. In many cases, governments need to make tough decisions and advance key reforms before the U.S. can help crowd in private capital. Many U.S. agencies provide technical assistance to strengthen policy and regulatory frameworks but lack concrete mechanisms to incentivize these reforms or make U.S. funding contingent on progress.
Limited tools supporting vital enabling public infrastructure blocks out private investment. The most challenging bottleneck to modernizing and strengthening a power sector is often not financing new power generation (which can easily attract private investment under the right conditions), but supporting critical enabling infrastructure including grid networks. In most emerging markets, these are public assets, wholly or partially state-owned. However, most U.S. energy finance tools are designed to support only private sector-led investments. This effectively limits their effectiveness to the generation sector, which already attracts far more capital than transmission or distribution.
To succeed, an energy security investment mechanism should:
- Enable investment highly tailored to the specific needs and priorities of partners;
- Provide support across the entire energy sector value chain, strengthening markets to enable greater direct investment by DFC and the private sector;
- Co-invest with partner countries in shared priorities, with strong accountability mechanisms.
Plan of Action
The new administration should work with Congress to give the Millennium Challenge Corporation the mandate to implement ‘Energy Security Compacts’ (ESCs) addressing the primary constraints to energy security in specific countries, and to coordinate the rest of the interagency in contributing relevant tools and resources. This proposal builds on and reflects key lessons learned from previous efforts by administrations of both parties.
Each Energy Security Compact would include the following:
- A process led by MCC and the National Security Council (NSC) to identify priority countries.
- An analysis jointly conducted by MCC and the partner country on the key constraints to energy security.
- Negotiation, led by MCC with support from NSC, of a multi-year Energy Security Compact, anchored by MCC support for a specific set of investments and reforms, and complemented by relevant contributions from the interagency. The Energy Security Compact would define agency-specific responsibilities and include clear objectives and measurable targets.
- Implementation of the Energy Security Compact, led by MCC and NSC. To manage this process, MCC and NSC would co-lead an Interagency Working Group comprising representatives from all relevant agencies.
- Results reporting, based on MCC’s top-ranked reporting process, to the National Security Council and Congress.
This would require the following congressional actions:
- Amend the Millennium Challenge Act of 2003: Grant MCC the expanded mandate to deploy Energy Security Compacts as a fourth business line. This should include language applying more flexible eligibility criteria to ESCs, and broadening the set of countries in which MCC can operate when implementing an ESC. Give MCC the mandate to co-lead an interagency working group with NSC.
- Plus up MCC Appropriation: ESCs can be launched as a pilot project in a few markets. But ultimately, the model’s success and impact will depend on MCC appropriations, including for direct investment and dedicated staff. MCC has a track record of outstanding transparency in evaluating its programs and reporting results.
- Strengthen DFC through reauthorization. The ultimate success of ESCs hinges on DFC’s ability to deploy more capital in the energy sector. DFC’s congressional authorization expires in September 2025, presenting an opportunity to enhance the agency’s reach and impact in energy security. Key recommendations for reauthorization include: 1) Addressing the equity scoring challenge; and 2) Raising DFC’s maximum contingent liability to $100 billion.
- Budget. The initiative could operate under various budget scenarios. The model is specifically designed to be scalable, based on the number of countries with which the U.S. wants to engage. It prioritizes efficiency by drawing on existing appropriations and authorities, by focusing U.S. resources on the highest priority countries and challenges, and by better coordinating the deployment of various U.S. tools.
This proposal draws heavily on the successes and struggles of initiatives from previous administrations of both parties. The most important lessons include:
- From MCC: The Compact model works. Multi-year Compact agreements are an effective way to ensure country buy-in, leadership, and accountability through the joint negotiation process and the establishment of clear goals and metrics. Compacts are also an effective mechanism to support hard infrastructure because they provide multi-year resources.
- From MCC: Investments should be based on rigorous analysis. MCC’s Constraints Analyses identify the most important constraints to economic growth in a given country. That same rigor should be applied to energy security, ensuring that U.S. investments target the highest impact projects, including those with the greatest positive impact on crowding in additional private sector capital.
- From Power Africa: Interagency coordination can work. Coordinating implementation across U.S. agencies is a chronic challenge. But it is essential to ESCs–and to successful energy investment more broadly. The ESC proposal draws on lessons learned from the Power Africa Coordinator’s Office. Specifically, joint-leadership with the NSC focuses effort and ensures alignment with broader strategic priorities. A mechanism to easily transfer funds from the Coordinator’s Office to other agencies incentivizes collaboration, and enables the U.S. to respond more quickly to unanticipated needs. And finally, staffing the office with individuals seconded from relevant agencies ensures that staff understand the available tools, how they can be deployed effectively, and how (and with whom) to work with to ensure success. Legislative language creating a Coordinator’s Office for ESCs can be modeled on language in the Electrify Africa Act of 2015, which created Power Africa’s interagency working group.
Conclusion
The new administration should work with Congress to empower the Millennium Challenge Corporation to lead the U.S. interagency in crafting ‘Energy Security Compacts’. This effort would provide the U.S. with the capability to coordinate direct investment in the energy security of a partner country and contribute to U.S. national priorities including diversifying energy supply chains, investing in the economic stability and performance of rapidly growing markets, and supporting allies with energy systems under direct threat.
This action-ready policy memo is part of Day One 2025 — our effort to bring forward bold policy ideas, grounded in science and evidence, that can tackle the country’s biggest challenges and bring us closer to the prosperous, equitable and safe future that we all hope for whoever takes office in 2025 and beyond.
PLEASE NOTE (February 2025): Since publication several government websites have been taken offline. We apologize for any broken links to once accessible public data.
MCC’s model already includes multi-year Compacts targeting major constraints to economic growth. The agency already has the structure and skills to implement Energy Security Compacts in place, including a strong track record of successful investment across many energy sector compacts. MCC enjoys a strong bipartisan reputation and consistently ranks as the world’s most transparent bilateral development donor. Finally, MCC is unique among U.S. agencies in being able to put large-scale grant capital into public infrastructure, a crucial tool for energy sector support–particularly in emerging and developing economies. Co-leading the design and implementation of ESCs with the NSC will ensure that MCC’s technical skills and experience are balanced with NSC’s view on strategic and diplomatic goals.
This proposal supports existing proposed legislative changes to increase MCC’s impact by expanding the set of countries eligible for support. The Millennium Challenge Act of 2003 currently defines the candidate country pool in a way that MCC has determined prevents it from “considering numerous middle-income countries that face substantial threats to their economic development paths and ability to reduce poverty.” Expanding that country pool would increase the potential for impact. Secondly, the country selection process for ESCs should be amended to include strategic considerations and to enable participation by the NSC.
America’s Teachers Innovate: A National Talent Surge for Teaching in the AI Era
Thanks to Melissa Moritz, Patricia Saenz-Armstrong, and Meghan Grady for their input on this memo.
Teaching our young children to be productive and engaged participants in our society and economy is, alongside national defense, the most essential job in our country. Yet the competitiveness and appeal of teaching in the United States has plummeted over the past decade. At least 55,000 teaching positions went unfilled this year, with long-term annual shortages set to double to 100,000 annually. Moreover, teachers have little confidence in their self-assessed ability to teach critical digital skills needed for an AI enabled future and in the profession at large. Efforts in economic peer countries such as Canada or China demonstrate that reversing this trend is feasible. The new Administration should announce a national talent surge to identify, scale, and recruit into innovative teacher preparation models, expand teacher leadership opportunities, and boost the profession’s prestige. “America’s Teachers Innovate” is an eight-part executive action plan to be coordinated by the White House Office of Science and Technology Policy (OSTP), with implementation support through GSA’s Challenge.Gov and accompanied by new competitive priorities in existing National Science Foundation (NSF), Department of Education (ED), Department of Labor (DoL), and Department of Defense education (DoDEA) programs.
Challenge and Opportunity
Artificial Intelligence may add an estimated $2.6 trillion to $4.4 trillion annually to the global economy. Yet, if the U.S. is not able to give its population the proper training to leverage these technologies effectively, the U.S. may witness a majority of this wealth flow to other countries over the next few decades while American workers are automated from, rather than empowered by, AI deployment within their sectors. The students who gain the digital, data, and AI foundations to work in tandem with these systems – currently only 5% of graduating high school students in the U.S. – will fare better in a modern job market than the majority who lack them. Among both countries and communities, the AI skills gap will supercharge existing digital divides and dramatically compound economic inequality.
China, India, Germany, Canada, and the U.K. have all made investments to dramatically reshape the student experience for the world of AI and train teachers to educate a modern, digitally-prepared workforce. While the U.S. made early research & development investments in computer science and data science education through the National Science Foundation, we have no teacher workforce ready to implement these innovations in curriculum or educational technology. The number of individuals completing a teacher preparation program has fallen 25% over the past decade; long-term forecasts suggest at least 100,000 shortages annually, teachers themselves are discouraging others from joining their own profession (especially in STEM), and preparing to teach digital skills such as computer science was the least popular option for prospective educators to pursue. In 2022, even Harvard discontinued its Undergraduate Teacher Education Program completely, citing low interest and enrollment numbers. There is still consistent evidence that young people or even current professionals remain interested in teaching as a possible career, but only if we create the conditions to translate that interest into action. U.S. policymakers have a narrow window to leverage the strong interest in AI to energize the education workforce, and ensure our future graduates are globally competitive for the digital frontier.
Plan of Action
America’s teaching profession needs a coordinated national strategy to reverse decades of decline and concurrently reinvigorate the sector for a new (and digital) industrial revolution now moving at an exponential pace. Key levers for this work include expanding the number of leadership opportunities for educators; identifying and scaling successful evidence-based models such as UTeach, residency-based programs, or National Writing Project’s peer-to-peer training sites; scaling registered apprenticeship programs or Grow Your Own programs along with the nation’s largest teacher colleges; and leveraging the platform of the President to boost recognition and prestige of the teaching profession.
The White House Office of Science and Technology Policy (OSTP) should coordinate a set of Executive Actions within the first 100 days of the next administration, including:
Recommendation 1. Launch a Grand Challenge for AI-Era Teacher Preparation
Create a national challenge via www.Challenge.Gov to identify the most innovative teacher recruitment, preparation, and training programs to prepare and retain educators for teaching in the era of AI. Challenge requirements should be minimal and flexible to encourage innovation, but could include the creation of teacher leadership opportunities, peer-network sites for professionals, and digital classroom resource exchanges. A challenge prompt could replicate the model of 100Kin10 or even leverage the existing network.
Recommendation 2. Update Areas of National Need
To enable existing scholarship programs to support AI readiness, the U.S. Department of Education should add “Artificial Intelligence,” “Data Science,” and “Machine Learning” to GAANN Areas of National Need under the Computer Science and Mathematics categories to expand eligibility for Masters-level scholarships for teachers to pursue additional study in these critical areas. The number of higher education programs in Data Science education has significantly increased in the past five years, with a small but increasing number of emerging Artificial Intelligence programs.
Recommendation 3. Expand and Simplify Key Programs for Technology-Focused Training
The President should direct the U.S. Secretary of Education, the National Science Foundation Director, and the Department of Defense Education Activity Director to add “Artificial Intelligence, Data Science, Computer Science” as competitive priorities where appropriate for existing grant or support programs that directly influence the national direction of teacher training and preparation, including the Teacher Quality Partnerships (ED) program, SEED (ED), the Hawkins Program (ED), the STEM Corps (NSF), the Robert Noyce Scholarship Program (NSF), and the DoDEA Professional Learning Division, and the Apprenticeship Building America grants from the U.S. Department of Labor. These terms could be added under prior “STEM” competitive priorities, such as the STEM Education Acts of 2014 and 2015 for “Computer Science,”and framed under “Digital Frontier Technologies.”
Additionally, the U.S. Department of Education should increase funding allocations for ESSA Evidence Tier-1 (“Demonstrates Rationale”), to expand the flexibility of existing grant programs to align with emerging technology proposals. As AI systems quickly update, few applicants have the opportunity to conduct rigorous evaluation studies or randomized control trials (RCTs) within the timespan of an ED grant program application window.
Additionally, the National Science Foundation should relaunch the 2014 Application Burden Taskforce to identify the greatest barriers in NSF application processes, update digital review infrastructure, review or modernize application criteria to recognize present-day technology realities, and set a 2-year deadline for recommendations to be implemented agency-wide. This ensures earlier-stage projects and non-traditional applicants (e.g. nonprofits, local education agencies, individual schools) can realistically pursue NSF funding. Recommendations may include a “tiered” approach for requirements based on grant size or applying institution.
Recommendation 4. Convene 100 Teacher Prep Programs for Action
The White House Office of Science & Technology Policy (OSTP) should host a national convening of nationally representative colleges of education and teacher preparation programs to 1) catalyze modernization efforts of program experiences and training content, and 2) develop recruitment strategies to revitalize interest in the teaching profession. A White House summit would help call attention to falling enrollment in teacher preparation programs; highlight innovative training models to recruit and retrain additional graduates; and create a deadline for states, districts, and private philanthropy to invest in teacher preparation programs. By leveraging the convening power of the White House, the Administration could make a profound impact on the teacher preparation ecosystem.
The administration should also consider announcing additional incentives or planning grants for regional or state-level teams in 1) catalyzing K-12 educator Registered Apprenticeship Program (RAPs) applications to the Department of Labor and 2) enabling teacher preparation program modernization for incorporating introductory computer science, data science, artificial intelligence, cybersecurity, and other “digital frontier skills,” via the grant programs in Recommendation 3 or via expanded eligibility for the Higher Education Act.
Recommendation 5. Launch a Digital “White House Data Science Fair”
Despite a bipartisan commitment to continue the annual White House Science Fair, the tradition ended in 2017. OSTP and the Committee on Science, Technology, and Math Education (Co-STEM) should resume the White House Science Fair and add a national “White House Data Science Fair,” a digital rendition of the Fair for the AI-era. K-12 and undergraduate student teams would have the opportunity to submit creative or customized applications of AI tools, machine-learning projects (similar to Kaggle competitions), applications of robotics, and data analysis projects centered on their own communities or global problems (climate change, global poverty, housing, etc.), under the mentorship of K-12 teachers. Similar to the original White House Science Fair, this recognition could draw from existing student competitions that have arisen over the past few years, including in Cleveland, Seattle, and nationally via AP Courses and out-of-school contexts. Partner Federal agencies should be encouraged to contribute their own educational resources and datasets through FC-STEM coordination, enabling students to work on a variety of topics across domains or interests (e.g. NASA, the U.S. Census, Bureau of Labor Statistics, etc.).
Recommendation 6. Announce a National Teacher Talent Surge at the State of Union
The President should launch a national teacher talent surge under the banner of “America’s Teachers Innovate,” a multi-agency communications campaign to reinvigorate the teaching profession and increase the number of teachers completing undergraduate or graduate degrees each year by 100,000. This announcement would follow the First 100 Days in office, allowing Recommendations 1-5 to be implemented and/or planned. The “America’s Teachers Innovate” campaign would include:
A national commitments campaign for investing in the future of American teaching, facilitated by the White House, involving State Education Agencies (SEAs) and Governors, the 100 largest school districts, industry, and philanthropy. Many U.S. education organizations are ready to take action. Commitments could include targeted scholarships to incentivize students to enter the profession, new grant programs for summer professional learning, and restructuring teacher payroll to become salaried annual jobs instead of nine-month compensation (see Discover Bank: “Surviving the Summer Paycheck Gap”).
Expansion of the Presidential Awards for Excellence in Mathematics and Science Teaching (PAMEST) program to include Data Science, Cybersecurity, AI, and other emerging technology areas, or a renaming of the program for wider eligibility across today’s STEM umbrella. Additionally, the PAMEST Award program should resume in-person award ceremonies beyond existing press releases, which were discontinued during COVID disruptions and have not since been offered. Several national STEM organizations and teacher associations have requested these events to return.
Student loan relief through the Teacher Loan Forgiveness (TLF) program for teachers who commit to five or more years in the classroom. New research suggests the lifetime return of college for education majors is near zero, only above a degree in Fine Arts. The administration should add “computer science, data science, and artificial intelligence” to the subject list of “Highly Qualified Teacher” who receive $17,500 of loan forgiveness via executive order.
An annual recruitment drive at college campus job fairs, facilitated directly under the banner of the White House Office of Science & Technology Policy (OSTP), to help grow awareness on the aforementioned programs directly with undergraduate students at formative career choice-points.
Recommendation 7. Direct IES and BLS to Support Teacher Shortage Forecasting Infrastructure
The IES Commissioner and BLS Commissioner should 1) establish a special joint task-force to better link existing Federal data across agencies and enable cross-state collaboration on the teacher workforce, 2) support state capacity-building for interoperable teacher workforce data systems through competitive grant priorities in the State Longitudinal Data Systems (SLDS) at IES and the Apprenticeship Building America (ABA) Program (Category 1 grants), and 3) recommend a review criteria question for education workforce data & forecasting in future EDA Tech Hub phases. The vast majority of states don’t currently have adequate data systems in place to track total demand (teacher vacancies), likely supply (teachers completing preparation programs), and the status of retention/mobility (teachers leaving the profession or relocating) based on near- or real-time information. Creating estimates for this very brief was challenging and subject to uncertainty. Without this visibility into the nuances of teacher supply, demand, and retention, school systems cannot accurately forecast and strategically fill classrooms.

Image: AmericanProgress.org
Recommendation 8. Direct the NSF to Expand Focus on Translating Evidence on AI Teaching to Schools and Districts.
The NSF Discovery Research PreK-12 Program Resource Center on Transformative Education Research and Translation (DRK-12 RC) program is intended to select intellectual partners as NSF seeks to enhance the overall influence and reach of the DRK-12 Program’s research and development investments. The DRK-12 RC program could be utilized to work with multi-sector constituencies to accelerate the identification and scaling of evidence-based practices for AI, data science, computer science, and other emerging tech fields. Currently, the program is anticipated to make only one single DRK-RC award; the program should be scaled to establish at least three centers: one for AI, integrated data science, and computer science, respectively, to ensure digitally-powered STEM education for all students.
Conclusion
China was #1 in the most recent Global Teacher Status Index, which measures the prestige, respect, and attractiveness of the teaching profession in a given country; meanwhile, the United States ranked just below Panama. The speed of AI means educational investments made by other countries have an exponential impact, and any misstep can place the United States far behind – if we aren’t already. Emerging digital threats from other major powers, increasing fluidity of talent and labor, and a remote-work economy makes our education system the primary lever to keep America competitive in a fast-changing global environment. The timing is ripe for a new Nation at Risk-level effort, if not an action on the scale of the original National Defense Education Act in 1958 or following the more recent America COMPETES Act. The next administration should take decisive action to rebuild our country’s teacher workforce and prepare our students for a future that may look very different from our current one.
This action-ready policy memo is part of Day One 2025 — our effort to bring forward bold policy ideas, grounded in science and evidence, that can tackle the country’s biggest challenges and bring us closer to the prosperous, equitable and safe future that we all hope for whoever takes office in 2025 and beyond.
PLEASE NOTE (February 2025): Since publication several government websites have been taken offline. We apologize for any broken links to once accessible public data.
This memo was developed in partnership with the Alliance for Learning Innovation, a coalition dedicated to advocating for building a better research and development infrastructure in education for the benefit of all students. Read more education R&D memos developed in partnership with ALI here.
Approximately 100,000 more per year. The U.S. has 3.2 million public school teachers and .5 million private school teachers (NCES, 2022). According to U.S. Department of Education data, 8% of public and 12% of private school teachers exit the profession each year (-316,000), a number that has remained relatively steady since 2012, while long-term estimates of re-entry continue to hover near 20% (+63,000). Unfortunately, the number of new teachers completing either traditional or alternative preparation programs has steadily declined over the past decade to 159,000+ per year. As a result of this gap, active vacancies continue to increase each year, and more than 270,000 educators are now cumulatively underqualified for their current roles, assumedly filling-in for absences caused by the widening gap. These predictions were made as early as 2016 (p. 2) and now have seemingly become a reality. Absent any changes, the total shortage of vacant or underqualified teaching positions could reach a total deficit between 700,000 and 1,000,000 by 2035.
The above shortage estimate assumes a base of 50,000 vacancies and 270,000 underqualified teachers as of the most recent available data, a flow of -94,000 net (entries – exits annually, including re-entrants) in 2023-2024. This range includes uncertainties for a slight (3%-5%) annual improvement in preparation from the status quo growth of alternative licensure pathways such as Grow your Own or apprenticeship programs through 2035. For exit rate, the most conservative estimates suggest a 5% exit rate, while the highest estimate at 50%; however, assembled state-level data suggests a 7.9% exit rate, similar to the NCES estimate (8%). Population forecasts for K-12 students (individuals aged 14-17) imply slight declines by 2035, based on U.S. Census estimates. Taken together, more optimistic assumptions result in a net cumulative shortage closer to -700,000 teachers, while worst-case scenario estimates may exceed -1,000,000.
Early versions of AI-powered tutoring have significant promise but have not yet lived up to expectations. Automated tutors have resulted in frustrating experiences for users, led students to perform worse on tests than those who leveraged no outside support, and have yet to successfully integrate other school subject problem areas (such as mathematics). We should expect AI tools to improve over time and become more additive for learning specific concepts, including repetitive or generalizable tasks requiring frequent practice, such as sentence writing or paragraph structure, which has the potential to make classroom time more useful and higher-impact. However, AI will struggle to replace other critical classroom needs inherent to young and middle-aged children, including classroom behavioral management, social motivation to learn, mentorship relationships, facilitating collaboration between students for project-based learning, and improving quality of work beyond accuracy or pre-prompted, rubric-based scoring. Teachers consistently report student interest as a top barrier for continued learning, which digital curriculum and AI automation may provide effectively for a short-period, but cannot do for the full twelve-year duration of a students’ K-12 experience.
These proposed executive actions complement a bi-partisan legislative proposal, “A National Training Program for AI-Ready Students,” which would invest in a national network of training sites for in-service teachers, provide grant dollars to support the expansion of teacher preparation programs, and help reset teacher payroll structure from 9-months to 12-months. Either proposal can be implemented independently from the other, but are stronger together.