This report was produced with generous support from Norwegian People’s Aid.

READ THE FULL REPORT HERE

The quantitative and qualitative enhancements to global nuclear arsenals in the past decade—particularly China’s nuclear buildup, Russia’s nuclear saber-rattling, and NATO’s response—have recently reinvigorated debates about how nuclear-armed states intend to use their nuclear weapons, and against which targets, in what some describe as a new Cold War. 

Details about who, what, where, when, why, and how countries target with their nuclear weapons are some of states’ most closely held secrets. Targeting information rarely reaches the public, and discussions almost exclusively take place behind closed doors—either in the depths of military headquarters and command posts, or in the halls of defense contractors and think tanks. The general public is, to a significant extent, excluded from those discussions. This is largely because nuclear weapons create unique expectations and requirements about secrecy and privileged access that, at times, can seem borderline undemocratic. Revealing targeting information could open up a country’s nuclear policies and intentions to intense scrutiny by its adversaries, its allies, and—crucially—its citizens. 

This presents a significant democratic challenge for nuclear-armed countries and the international community. Despite the profound implications for national and international security, the intense secrecy means that most individuals—not only including the citizens of nuclear-armed countries and others that would bear the consequences of nuclear use, but also lawmakers in nuclear-armed and nuclear umbrella states that vote on nuclear weapons programs and policies—do not have much understanding of how countries make fateful decisions about what to target during wartime, and how. When lawmakers in nuclear-armed countries approve military spending bills that enhance or increase nuclear and conventional forces, they often do so with little knowledge of how those bills could have implications for nuclear targeting plans. And individuals across the globe do not know whether they live in places that are likely to be nuclear targets, or what the consequences of a nuclear war would be.

While it is reasonable for governments to keep the most sensitive aspects of nuclear policies secret, the rights of their citizens to have access to general knowledge about these issues is equally valid so they may know about the consequences to themselves and their country, and so that they can make informed assessments and decisions about their respective government’s nuclear policies. Under ideal conditions, individuals should reasonably be able to know whether their cities or nearby military bases are nuclear targets and whether their government’s policies make it more or less likely that nuclear weapons will be used.

As an organization that seeks to empower individuals, lawmakers, and journalists with factual information about critical topics that most affect them, the Federation of American Scientists—through this report—aims to help fill some of these significant knowledge gaps. This report illuminates what we know and do not know about each country’s nuclear targeting policies and practices, and considers how they are formulated, how they have changed in recent decades, whether allies play a role in influencing them, and why some countries are more open about their policies than others. The report does not claim to be comprehensive or complete, but rather should be considered as a primer to help inform the public, policymakers, and other stakeholders. This report may be updated as more information becomes available.

Given the secrecy associated with nuclear targeting information, it is important at the outset to acknowledge the limitations of using exclusively open sources to conduct analysis on this topic. Information in and about different nuclear-armed states varies significantly. For countries like the United States—where nuclear targeting policies have been publicly described and are regularly debated inside and outside of government among subject matter experts—official sources can be used to obtain a basic understanding of how nuclear targets are nominated, vetted, and ultimately selected, as well as how targeting fits into the military strategy. However, there is very little publicly available information about the nuclear strike plans themselves or the specific methodology and assumptions that underpin them. For less transparent countries like Russia and China—where targeting strategy and plans are rarely discussed in public—media sources, third-country intelligence estimates, and nuclear force structure analysis can be used, in conjunction with official statements or statements from retired officials, to make educated assumptions about targeting policies and strategies. 

It is important to note that a country’s relative level of transparency regarding its nuclear targeting policies does not necessarily echo its level of transparency regarding other aspects of its governance structure. Ironically, some of the most secretive and authoritarian nuclear-armed states are remarkably vocal about what they would target in a nuclear war. This is typically because those same countries use nuclear rhetoric as a means to communicate deterrence signals to their respective adversaries and to demonstrate to their own population that they are standing up to foreign threats. For example, while North Korea keeps many aspects of its nuclear program secret, it has occasionally stated precisely which high-profile targets in South Korea and across the Indo-Pacific region it would strike with nuclear weapons. In contrast, some other countries might consider that frequently issuing nuclear threats or openly discussing targeting policies could potentially undermine their strategic deterrent and even lower the threshold for nuclear use.

READ THE FULL REPORT HERE

Nearly one year after the Pentagon certified the Sentinel intercontinental ballistic missile program to continue after it incurred critical cost and schedule overruns, the new nuclear missile could once again be in trouble.¹

An April 16th article from Defense Daily broke the news that the Air Force will have to dig new holes for the Sentinel silos.² The service had been planning to refurbish the existing 450 Minuteman silos but recently discovered, as noted in a follow-up article from Breaking Defense, that the silos will “largely not be reusable after all.”³ Brig. Gen. William Rogers, the Air Force’s director of the ICBM Systems Directorate, cited asbestos, lead paint, and other issues with the existing silos that make refurbishment difficult.⁴ Air Force officials also stated that an ongoing study into missileer cancer rates played a role in the decision to build new silos.⁵ 

This news comes shortly after reports that the Air Force is planning to extend the life of the currently deployed Minuteman III ICBMs until “at least” 2050—roughly 20 years beyond their intended service lives—due to delays in the Sentinel program.⁶ 

For those who have been tracking the Sentinel development since the Air Force first conceptualized a new ICBM in the early 2010s, the reports of Minuteman life-extension likely made them pause and recall the common refrain from Sentinel proponents over the years that life-extending Minuteman III missiles would be too expensive or even impossible. “You cannot life-extend Minuteman III,” then-commander of US Strategic Command Adm. Charles Richard told reporters in 2021.⁷ In 2016, the Air Force told Congress that the Minuteman III was aging out, therefore the “GBSD solution” was necessary to ensure the future viability of the ICBM force (GBSD is short for Ground-Based Strategic Deterrent, the programmatic name for the ICBM before Sentinel was chosen in 2022). Air Force officials still maintain that a life-extension program for Minuteman is not possible. In their words, Minuteman will be “sustain[ed] to keep it viable until Sentinel is delivered.”⁸ Regardless of how the Air Force refers to the effort, it appears that Minuteman III will be made to operate well beyond its planned service life.

For some, like our team at the Federation of American Scientists’ Nuclear Information Project, Sentinel’s newest struggles came as no surprise at all. For years, it has been clear to observers that this program has suffered from chronic unaccountability, overconfidence, poor performance, and mismanagement. Project benchmarks were cherry-picked, viable alternatives were prematurely dismissed, competition was discouraged, and goalposts were continuously moved. Ultimately, it will be U.S. taxpayers who pay the increasingly rising costs, and other—more critical—priorities will suffer as Sentinel continuously sucks money away from other programs. 

It comes as no surprise that Sentinel was specifically named in the White House’s recent memo requiring all Major Defense Acquisition Programs more than 15% over-budget or behind schedule to be “reviewed for cancellation;” Sentinel is the poster-child for inefficiency, which the administration claims to be obsessed with eliminating.⁹ In order to prevent this type of mismanagement for future programs, we must first understand how Sentinel went so wrong. 

How We Got Here

The Federation of American Scientists has been intensively tracking the progress of the Sentinel program for years. Throughout the acquisition process, the Air Force clung to its fundamental and counterintuitive assumption that building an entirely new ICBM from scratch would be cheaper than life-extending the current system. We now know that this assumption was wildly incorrect, but how did it reach this point? 

Cherry-picked project benchmarks

When seeking to plug a capability gap, the Pentagon is required to consider a range of procurement options before proceeding with its acquisition. This process takes place over several years and culminates in an “Analysis of Alternatives”—a comparative evaluation of the operational effectiveness, suitability, risk, and life-cycle costs of the various options under consideration. This assessment can have tremendous implications for an acquisition program, as it documents the rationale for recommending a particular course of action. 

The Air Force’s Analysis of Alternatives for the program that would eventually become Sentinel was conducted between 2013 and 2014, and concluded that the costs of pursuing a Minuteman III life-extension would be nearly the same as those projected for Sentinel.¹⁰ Crucially, this cost comparison was pegged to a predetermined requirement to continue deploying the same number of missiles until the year 2075.¹¹ 

These benchmarks, despite having no apparent inalterable national security imperative, appear to have played a significant role in shaping perceptions of the two options. While it is now clear that Minuteman III could be—and likely will be—life-extended for several more decades, the Air Force does not have enough airframes to keep at least 400 of them in service through 2075 and maintain the testing campaign needed to ensure reliability. As a result, in order to push the ICBM force beyond 2075, the Air Force would need to life-extend Minuteman III and pursue a follow-on system after that point. 

This was reportedly reflected in the Air Force’s cost analysis, which explains why the cost of the Minuteman III life-extension option was estimated by the Air Force to be roughly the same as the cost of building an entirely new ICBM.¹² The service was not simply comparing the costs of a life-extension and a brand-new system; it was instead comparing the costs of pursuing Sentinel immediately on the one hand, versus a Minuteman III life-extension and development of a follow-on system on the other hand. 

Of course, policymakers require benchmarks in order to make estimates: it would not be reasonable to analyze the feasibility of a particular system without considering how long and at what level that system needs to perform. However, in the case of the Sentinel, selecting those particular benchmarks at the beginning of the process essentially pre-baked the analysis before it even began in earnest. 

Let’s say a different evaluation benchmark had been selected—2050, for example, rather than 2075. 

In January 2025, Defense Daily reported that the Air Force would likely have to keep portions of the Minuteman III fleet in service until 2050 or later.¹³ This may require altering certain aspects of the Minuteman III’s deployment—such as reducing the number of deployed ICBMs or annual test launches in order to preserve airframes. While no final decisions have been made, the Air Force is clearly evaluating continued reliance on Minuteman III as a potential option, despite years of high-ranking military and political officials stating that doing so was impossible.¹⁴ 

Benchmarking the cost analysis at 2050 rather than 2075 would have thus yielded wildly different results. In 2012, the Air Force admitted that it cost only $7 billion to modernize its Minuteman III ICBMs into “basically new missiles except for the shell.”¹⁵ While getting those same missiles past 2050 would certainly add additional cost and complexity—particularly to replace parts whose manufacturers no longer exist—it is unfathomable that the costs would come anywhere close to those of the Sentinel program, which was estimated by the Pentagon’s Director of Cost Assessment and Program Evaluation (CAPE) in 2020 (before the critical cost overrun) to have a total lifecycle cost of $264 billion in then-year dollars.

It is particularly troubling that very few public or independent government-sponsored analyses were conducted to look into the Sentinel program’s flawed assumptions, nor the realistic possibility of a Minuteman III life-extension. Countless congressional and non-governmental attempts to push for one were stymied at every turn. In 2019, for example, dozens of lobbyists from the Sentinel contract bidders successfully helped to eliminate a proposed amendment to the National Defense Authorization Act calling for an independent study on a Minuteman III life-extension program.¹⁶ 

The most comprehensive public study on this issue was a 2022 report published by the Carnegie Endowment for International Peace under contract from the Pentagon; however, the study noted that “the iterative process through which we received information, the unclassified nature of our study, and the limited time available for investigating DOD conclusions left us unable to assess the DOD’s position regarding the technical and cost feasibility of an extended Minuteman III alternative to GBSD;” the authors ultimately concluded that a more detailed technical analysis was required in order to answer these questions.¹⁷ 

While the findings of such a study will never be known, it is likely that they would have supported what was clear to government watchdogs at the time and has been validated in spades since then: the assumptions baked into this program were flawed from the start, and the system’s costs would be significantly larger than initially expected. Given that the Pentagon ultimately went in the opposite direction, taxpayers are now on the hook for both a de facto Minuteman III life-extension program as well as the substantial costs associated with acquiring Sentinel—with limited further possibilities for near-term cost mitigation. 

Failure to predict the true costs and needs of the program

In addition to the cherry-picked benchmarks that tipped the scales towards a brand-new ICBM, when comparing costs the Air Force made a key error in its assumptions: it assumed that the Sentinel would be able to reuse much of the original Minuteman launch infrastructure. 

Some level of infrastructure modernization for the Sentinel was always planned, including building entirely new launch control centers and additional infrastructure for the launch facilities.¹⁸ However, the original plans called for reusing existing copper command and control cabling and the refurbishment—not reconstruction—of 450 silos. Both assumptions have proven incorrect, and perhaps more than anything else, now represent the single greatest driver of Sentinel’s skyrocketing costs. 

While both the current cabling and launch facilities work fine for the existing Minuteman III and would presumably function similarly following a life-extension, they are apparently incompatible with Sentinel’s increasingly complex design. 

The Air Force must now dig up and replace 7,500 miles of cabling with the latest fiber optic cables. Much of these cables are buried underneath private property, meaning that local landowners must lease 100-foot-wide lines on their property to the Pentagon to be dug up for multi-year periods.¹⁹ 

In addition, both the Air Force and Northrop Grumman have now recognized that it will take more than simple refurbishments to make the existing Minuteman III launch facilities compatible with Sentinel. Both the service and the contractor have stated that several of the assumptions regarding the conversion process that went into the 2020 baseline review have now proved to be incorrect.²⁰ 

As a result, the Air Force is apparently now planning to build entirely new launch facilities to house the Sentinel, most of which will require digging new holes in the ground.²¹ As one Northrop Grumman official explained, “When you multiply that by 450, if every silo is a little bit bigger or has an extra component, that actually drives a lot of cost because of the sheer number of them that are being updated.”²² It is unclear whether the costs will increase beyond the new estimate released with the Nunn-McCurdy decision, but the program is clearly trending in the wrong direction.

The Air Force had been publicly teasing the prospect of digging new holes for nearly a year. At the Triad Symposium in Washington, D.C., in September 2024, Maj. Gen. Colin Connor, director of ICBM Modernization at Barksdale Air Force Base, responded to an audience question about the new silos rumor by saying, “we’re looking at all of our options.” Despite the noncommittal answer, the decision to dig new silos seems to have already been made by the time of Connor’s statement. 

Firstly, it has since been revealed that the estimated costs of the new silos were included in the Nunn-McCurdy review process which concluded in July 2024. Additionally, although the decision was not made public until the April 16 Defense News article, Northrop Grumman may have inadvertently revealed the news much earlier. Included in the gallery of images of the Sentinel program on Northrop’s website is a digital mockup of a Sentinel launch facility. The first version of the image (see Figure A below) illustrates the Air Force’s original plan to refurbish the Minuteman III silos for Sentinel, with a key indicating the silo and silo lid as “Reclaimed MMIII Facilities.” A newer version of the image (see Figure B below) was uploaded to the gallery as early as February 2024 and shows the entire launch facility—including the silo and silo lid—as “New Sentinel Facilities.”

Figure A

Original rendering of Sentinel launch facility. (Source: Northrop Grumman)

Figure B

New rendering of Sentinel launch facility. (Source: Northrop Grumman)

Unwarranted overconfidence 

Despite the clear concerns outlined above, the Pentagon was remarkably confident in its and Northrop Grumman’s abilities to deliver the Sentinel on-time and on-budget. 

In September 2020, the Pentagon delivered its Milestone B summary report to Congress—a key decision point at which acquisition programs are authorized to enter the Engineering and Manufacturing Development phase, considered to be the official start of a program. The Milestone B report included an estimate of $95.8 billion in then-year dollars to acquire the Sentinel—a significant increase from previous estimates, but not yet the dire situation that we find ourselves in today (Figure C). 

Figure C

The above table from the Congressional Budget Office shows the cost growth for the Sentinel’s acquisition program between the Sentinel’s Milestone B assessment in 2020 and the post-Nunn-McCurdy review process in 2025. All costs are reflected in FY2020 dollars to allow for an accurate comparison between years.

We now know, however, based on recent statements from Pentagon and Air Force officials, that there were “some gaps in maturity” in the Milestone B report.²³ Specifically, “in September of 2020, the knowledge of the ground-based segment of this program was insufficient in hindsight to have a high-quality cost estimate.” What this means is that at the most consequential stage of the program to-date, it was approved without a comprehensive understanding of the likely cost growth. 

Furthermore, the Air Force was heavily delayed in creating an integrated master schedule for the Sentinel program. An integrated master schedule includes the planned work, the resources necessary to accomplish that work, and the associated budget; from the government’s perspective, it is considered to be the keystone for program management.²⁴ Although the Under Secretary of Defense for Acquisition and Sustainment testified to Congress that “By the time you’re six months after Milestone B, you should have an integrated master schedule,” the Air Force had not met this mark.²⁵ If the Air Force did manage to create such a schedule, it became obsolete with the Nunn-McCurdy Act’s requirement to restructure the program and rescind its Milestone B approval.

During that same hearing, the Air Force’s Deputy Chief of Staff for Strategic Deterrence and Nuclear Integration also admitted that at that time, the service had been experiencing poor communication with Northrop Grumman, the primary contractor for the ICBM. 

Performance issues also appear to have had an impact on the program. In June 2024, the Air Force removed the colonel in charge of its Sentinel program—reportedly for a “failure to follow operational procedures”—and replaced him with a two-star general, with the rank change indicating a need for greater high-level attention.²⁶ 

Throughout this time, the Air Force remained overconfident in its abilities to deliver the program; in December 2020, the Assistant Secretary of the Air Force for Acquisition, Technology, and Logistics told reporters that the Air Force had “godlike insight into all things GBSD.”²⁷ And in September 2022, the Air Force Major General responsible for Sentinel’s strategic planning and requirements said in a Breaking Defense interview that the program was “on cost, on schedule, and the acquisition program baseline is being met.”²⁸ 

Given everything we now know about the state of the Sentinel program, these statements were either clear obfuscations or just pure fantasy. 

Non-competitive disadvantages

When addressing concerns about the rising projected costs of the Sentinel program, Air Force leaders were confident that a competitive and healthy industrial base would be able to keep the overall price tag down. As Gen. Timothy Ray, then-Commander of Air Force Global Strike Command, told reporters in 2019, “our estimates are in the billions of savings over the lifespan of the weapon.”²⁹ 

These expected savings clearly never materialized, however, nor did the Pentagon help facilitate the conditions for them to be realized. In March 2018, the Air Force Nuclear Weapons Center submitted a document justifying its intention to restrict competition for the Sentinel contract to just two suppliers—Boeing and Northrop Grumman—stating that this limitation would still constrain costs because the two companies would be in competition with one another.³⁰ 

However, this specter of competition evaporated when Boeing withdrew from the competition following Northrop Grumman’s acquisition of Orbital ATK—one of two independent producers of large solid rocket motors left in the US market.³¹ As these motors are necessary to make ICBMs fly, the merger put Northrop Grumman in the driver’s seat: it could restrict access to those motors from Boeing, thus tanking its competitor’s chances at the Sentinel bid. 

Doing so would not have been allowed by the terms of the Federal Trade Commission, which permitted the merger in 2018 but subsequently investigated it in 2022 under the Biden administration, and also subsequently blocked a similar attempted merger between Lockheed Martin and Aerojet Rocketdyne that same year.³² However, the Pentagon, which had initially included non-exclusionary and pro-competition language in its requirements for an earlier phase of the Sentinel contract, removed that language from future phases.³³ By refusing to wield its own power to preserve competition—initially a key driver for promoting Sentinel over a Minuteman III life-extension—the Air Force essentially left the state of the competition in Northrop Grumman’s hands. According to Boeing’s CEO, Northrop Grumman subsequently slow-walked the process of hammering out a competition arrangement with Boeing—apparently not leaving enough time for Boeing to negotiate a competitive price for solid rocket motors before the Sentinel deadline.³⁴ 

As a result, Boeing pulled out of the competition altogether, and the Air Force awarded the Sentinel engineering and manufacturing development contract to Northrop Grumman through an unprecedented single-source bidding process. As the Under Secretary of Defense for Acquisition and Sustainment admitted during 2024 testimony to Congress, what this amounted to was that “effectively there was not, at the end of the day, competition in this program.”³⁵ 

Reflecting on the Sentinel procurement process, House Armed Services Committee chairman Adam Smith—who has a sizable Boeing presence in his home state of Washington—suggested in October 2019 that the Air Force is “way too close to the contractors they are working with,” and implied that the service was biased towards Northrop Grumman.³⁶ 

Predictably, the evaporation of competition has coincided with skyrocketing Sentinel acquisition costs. In July 2024, the Air Force’s acquisition chief Andrew Hunter reportedly told reporters that the Air Force was considering reopening parts of the Sentinel contract to bids. “I think there are elements of the ground infrastructure where there may be opportunities for competition that we can add to the acquisition strategy for Sentinel,” Hunter said.³⁷

The Nunn-McCurdy Saga

In January 2024, the Air Force notified Congress that the Sentinel program had incurred a critical breach of the Nunn-McCurdy Act, legislation designed to keep expensive programs in check.³⁸ One week after notifying Congress of the breach, the Air Force fired the head of the Sentinel program, but said the move was “not directly related” to the Nunn-McCurdy breach.³⁹

At the time of the notification, the Air Force stated that the program was 37% over budget and two years behind schedule. Six months later, after conducting the cost reassessment mandated by Nunn-McCurdy, the Pentagon announced that the Sentinel program would cost 81% more than projected and be delayed by several years.⁴⁰ Nevertheless, the Secretary of Defense certified the program to continue. 

Per the requirements of the Nunn-McCurdy Act, the Under Secretary of Defense for Acquisition and Sustainment, who serves as the Milestone Decision Authority for the program, rescinded Sentinel’s Milestone B approval, which is needed for a program to enter the engineering and manufacturing development phase.⁴¹ The Air Force must restructure the program to address the root cause of the cost growth before receiving a new milestone approval, a process the service has said will take approximately 18 to 24 months.⁴² 

Where Sentinel Stands Now

Work on the Sentinel program has continued while the Air Force carries out the restructuring effort, but the government can’t seem to decide whether things are going well or not.  

On February 10, the Air Force told Defense One that parts of the Sentinel program had been “suspended.”⁴³ Due to “evolving” requirements related to Sentinel launch facilities, the Air Force instructed Northrop Grumman to halt “design, testing, and construction work related to the Command & Launch Segment.” There has been no indication of when the stop work order will be lifted. Nevertheless, during an April 10 Air Force town hall on Sentinel in Kimball, Nebraska, Wing Commander of F.E. Warren AFB Col. Johnny Galbert told attendees that Sentinel “is not on hold; it is moving forward.”⁴⁴

Just under one month after the stop work order was made, the Air Force announced that the Sentinel program had achieved a “modernization milestone” with the successful static fire test of Sentinel’s stage-one solid rocket motor.⁴⁵ The test marked the successful test firing of each stage of Sentinel’s rocket motor after the second and third stages were tested in 2024. 

On March 27, the same day Bloomberg reported that the Air Force was considering a life-extension program for Minuteman III missiles, President Trump’s nominee for Secretary of the Air Force (confirmed by the Senate on May 13), Troy Meinke, committed in his testimony to pushing Sentinel over the finish line, calling the program “foundational to strategic deterrence and defense of the homeland.”⁴⁶ During the same hearing, Trump’s nominee for undersecretary of Defense for acquisition and sustainment, Michael P. Duffey, also shared his support for the Sentinel program, saying “nuclear modernization is the backbone of our strategic deterrent,” and endorsing Sentinel as “critical.” Yet, two weeks later, on April 9, President Trump signed an executive order to address defense acquisition programs that mandates, “any program more than 15% behind schedule or 15% over cost will be scrutinized for cancellation.”⁴⁷ This places Sentinel well beyond the threshold for potential cancellation, and the White House fact sheet detailing the order explicitly called out Sentinel’s cost and schedule overruns.

The next day, the Air Force announced that another “key milestone” for the Sentinel program had been met with the stand-up of Detachment 11 at Malmstrom AFB, which will oversee implementation of the Sentinel program at the base.⁴⁸ But of course, less than thirty days later, Sentinel took a major blow with the Air Force’s admittance that hundreds of new silos would have to be dug up and constructed for the new ICBM. 

The Government Accountability Office’s (GAO) latest Weapon Systems Annual Assessment from June 11 reports that Sentinel’s costs “could swell further” as the Air Force “continues to evaluate its options and develop a new schedule as part of restructuring efforts.” The assessment also notes that the Sentinel program alone accounted for over $36 billion of the $49.3 billion increase from 2024 to 2025 in GAO’s combined total estimate of major defense acquisition program costs, and noted that the first flight test now would not take place until March 2028.⁴⁹ In a sweeping criticism of the program, the GAO report notes that the continued immaturity of the program’s critical technologies more than 4 years into its development phase “calls into question the level of work required to mature these technologies and the validity of the cost estimate used to certify the program.”⁵⁰

450 Money Pits

We probably will never know how much money could have been saved if the Air Force had elected from the beginning to life-extend the existing ICBMs rather than build an entirely new system from scratch. The opportunity to have a proactive, independent cost comparison and corresponding public debate was eliminated through intense rounds of Pentagon and industry lobbying. But we certainly now know that the Air Force’s assertion—that the Sentinel would be cheaper and easier than a life-extension—was wrong, and that the suppression of an independent review contributed to these rising costs. 

The Sentinel saga, with its seemingly unending series of setbacks and continued uncertainties, begs a crucial question: what incentives exist for the Air Force to get it right? That the program, along with numerous other nuclear modernization programs, was green-lighted to continue despite ever-increasing cost and schedule delays exposes a major flaw in U.S. nuclear weapons acquisition programs – they are too big to fail. The government, evidently, will always write a bigger check, will always move the goalposts, because the alternative is either failing to maintain the U.S. strategic deterrent or admitting that U.S. nuclear strategy and force structure is not as immutable and unquestionable as the public has been made to believe. In such a system of blank checks and industry lobbying, what incentivizes the Pentagon to ensure programs are as cost efficient as possible? The only mechanism for oversight and accountability is Congress. Congress must increase oversight of nuclear modernization programs like Sentinel to ensure a limit is placed on how much taxpayer money can be spent on failing programs in the name of national security.

The Stockholm International Peace Research Institute (SIPRI) launches its annual assessment of the state of armaments, disarmament and international security

Washington, D.C. – June 16, 2025 – Nuclear weapons researchers at the Federation of American Scientists (FAS) contributed to the Stockholm International Peace Research Institute (SIPRI)’s annual Yearbook, released today. Key findings of SIPRI Yearbook 2025 are that a dangerous new nuclear arms race is emerging at a time when arms control regimes are severely weakened.

“The era of reductions in the number of nuclear weapons in the world, which had lasted since the end of the cold war, is coming to an end,” said Hans M. Kristensen, Associate Senior Fellow with SIPRI’s Weapons of Mass Destruction Programme and Director of the Nuclear Information Project at the Federation of American Scientists (FAS). “Instead, we see a clear trend of growing nuclear arsenals, sharpened nuclear rhetoric and the abandonment of arms control agreements.”

World’s nuclear arsenals being enlarged and upgraded  

Nearly all of the nine nuclear-armed states—the United States, Russia, the United Kingdom, France, China, India, Pakistan, the Democratic People’s Republic of Korea (North Korea) and Israel—continued intensive nuclear modernization programmes in 2024, upgrading existing  weapons and adding newer versions. 

Of the total global inventory of an estimated 12,241 warheads in January 2025, about 9614 were in military stockpiles for potential use (see the table below). An estimated 3912 of those warheads were deployed with missiles and aircraft and the rest were in central storage. Around 2100 of the deployed warheads were kept in a state of high operational alert on ballistic missiles. Nearly all of these warheads belonged to Russia or the USA, but China may now keep some warheads on missiles during peacetime. 

Since the end of the cold war, the gradual dismantlement of retired warheads by Russia and the USA has normally outstripped the deployment of new warheads, resulting in an overall year-on year decrease in the global inventory of nuclear weapons. This trend is likely to be reversed in the coming years, as the pace of dismantlement is slowing, while the deployment of new nuclear weapons is accelerating. 

Russia and the USA together possess around 90 per cent of all nuclear weapons. The sizes of their respective military stockpiles (i.e. useable warheads) seem to have stayed relatively stable in 2024 but both states are implementing extensive modernization programmes that could increase the size and diversity of their arsenals in the future. If no new agreement is reached to cap their stockpiles, the number of warheads they deploy on strategic missiles seems likely to increase after the bilateral 2010 Treaty on Measures for the Further Reduction and Limitation of Strategic Offensive Arms (New START) expires in February 2026. 

The USA’s comprehensive nuclear modernization programme is progressing but in 2024 faced planning and funding challenges that could delay and significantly increase the cost of the new strategic arsenal. Moreover, the addition of new non-strategic nuclear weapons to the US arsenal will place further stress on the modernization programme. 

Russia’s nuclear modernization programme is also facing challenges that in 2024 included a test  failure and the further delay of the new Sarmat intercontinental ballistic missile (ICBM) and slower  than expected upgrades of other systems. Furthermore, an increase in Russia’s non-strategic nuclear warheads predicted by the USA in 2020 has so far not materialized. 

###

ABOUT FAS

The Federation of American Scientists (FAS) works to advance progress on a broad suite of contemporary issues where science, technology, and innovation policy can deliver transformative impact, and seeks to ensure that scientific and technical expertise have a seat at the policymaking table. Established in 1945 by scientists in response to the atomic bomb, FAS continues to bring scientific rigor and analysis to address contemporary challenges. More information about FAS work at fas.org.

As the United States overhauls nearly every element of its strategic nuclear forces, artificial intelligence is set to play a larger role—initially in early‑warning sensors and decision‑support tools, and likely in other mission areas. Improved detection could strengthen deterrence, but only if accompanying hazards—automation bias, model hallucinations, exploitable software vulnerabilities, and the risk of eroding assured second‑strike capability—are well managed. 

To ensure responsible AI integration, the Office of the Assistant Secretary of Defense for Nuclear Deterrence, Chemical, and Biological Defense Policy and Programs (OASD (ND-CBD)), the U.S. Strategic Command (STRATCOM), the Defense Advanced Research Projects Agency (DARPA), the Office of the Undersecretary of Defense for Policy (OUSD(P)), and the National Nuclear Security Administration (NNSA), should jointly develop a standardized AI risk-assessment framework guidance document, with implementation led by the Department of Defense’s Chief Digital and Artificial Intelligence Office (CDAO) and STRATCOM. Furthermore, DARPA and CDAO should join the Nuclear Weapons Council to ensure AI-related risks are systematically evaluated alongside traditional nuclear modernization decisions. 

Challenge and Opportunity 

The United States is replacing or modernizing nearly every component of its strategic nuclear forces, estimated to cost at least $1.7 trillion over the next 30 years. This includes its:

• Intercontinental ballistic missiles (ICBMs)
• Ballistic missile submarines and their submarine-launched ballistic missiles (SLBMs) 
• Strategic bombers, cruise missiles, and gravity bombs
• Nuclear warhead production and plutonium pit fabrication facilities 

Simultaneously, artificial intelligence (AI) capabilities are rapidly advancing and being applied across the national security enterprise, including nuclear weapons stockpile stewardship and some components of command, control, and communications (NC3) systems, which encompass early warning, decision-making, and force deployment components.

The NNSA, responsible for stockpile stewardship, is increasingly integrating AI into its work. This includes using AI for advanced modeling and simulation of nuclear warheads. For example, by creating a digital twin of existing weapons systems to analyze aging and performance issues, as well as using AI to accelerate the lifecycle of nuclear weapons development. Furthermore, NNSA is leading some aspects of the safety testing and systematic evaluations of frontier AI models on behalf of the U.S. government, with a specific focus on assessing nuclear and radiological risk.

Within the NC3 architecture, a complex “system of systems” with over 200 components, simpler forms of AI are already being used in areas including early‑warning sensors, and may be applied to  decision‑support tools and other subsystems as confidence and capability grow. General Anthony J. Cotton—who leads STRATCOM, the combatant command that directs America’s global nuclear forces and their command‑and‑control network—told a 2024 conference that STRATCOM is “exploring all possible technologies, techniques, and methods” to modernize NC3. Advanced AI and data‑analytics tools, he said, can sharpen decision‑making, fuse nuclear and conventional operations, speed data‑sharing with allies, and thus strengthen deterrence. General Cotton added that research must also map the cascading risks, emergent behaviors, and unintended pathways that AI could introduce into nuclear decision processes.

Thus, from stockpile stewardship to NC3 systems, AI is likely to be integrated across multiple nuclear capabilities, some potentially stabilizing, others potentially highly destabilizing. For example, on the stabilizing effects, AI could enhance early warning systems by processing large volumes of satellite, radar, and other signals intelligence, thus providing more time to decision-makers. On the destabilizing side, the ability for AI to detect or track other countries’ nuclear forces could be destabilizing, triggering an expansionary arms race if countries doubt the credibility of their second-strike capability. Furthermore, countries may misinterpret each other’s nuclear deterrence doctrines or have no means of verification of human control of their nuclear weapons.

While several public research reports have been conducted on how AI integration into NC3 could upset the balance of strategic stability, less research has focused on the fundamental challenges with AI systems themselves that must be accounted for in any risk framework. Per the National Institute of Standards and Technology’s (NIST) AI Risk Management Framework, several fundamental AI challenges at a technical level must be accounted for in the integration of AI into stockpile stewardship and NC3.

Not all AI applications within the nuclear enterprise carry the same level of risk. For example, using AI to model warhead aging in stockpile stewardship is largely internal to the Department of Energy (DOE) and involves less operational risk. Despite lower risk, there is still potential for an insufficiently secure model to lead to leaked technical data about nuclear weapons. 

However, integrating AI into decision support systems or early warning functions within NC3 introduces significantly higher stakes. These systems require time-sensitive, high-consequence judgments, and AI integration in this context raises serious concerns about issues including confabulations, human-AI interactions, and information security: 

• Confabulations: A phenomenon in which generative AI systems (GAI) systems generate and confidently present erroneous or false content in response to user inputs, or
  prompts. These phenomena are colloquially also referred to as “hallucinations” or “fabrications”, and could have particularly dangerous consequences in high-stakes settings. 

• Human-AI Interactions: Due to the complexity and human-like nature of GAI technology, humans may over-rely on GAI systems or may unjustifiably perceive GAI content to be of higher quality than that produced by other sources. This phenomenon is an example of automation bias or excessive deference to automated systems. This deference can lead to a shift from a human making the final decision (“human in the loop”), to a human merely observing AI generated decisions (“human on the loop”). Automation bias therefore risks exacerbating other risks of GAI systems as it can lead to humans maintaining insufficient oversight.

• Information Security: AI expands the cyberattack surface of NC3. Poisoned AI training data and tampered code can embed backdoors, and, once deployed, prompt‑injection or adversarial examples can hijack AI decision tools, distort early‑warning analytics, or leak secret data. The opacity of large AI models can let these exploits spread unnoticed, and as models become more complex, they will be harder to debug.

This is not an exhaustive list of issues with AI systems, however it highlights several key areas that must be managed. A risk framework must account for these distinctions and apply stricter oversight where system failure could have direct consequences for escalation or deterrence credibility. Without such a framework, it will be challenging to harness the benefits AI has to offer.

Plan of Action 

Recommendation 1. OASD (ND-CBD),  STRATCOM, DARPA, OUSD(P), and NNSA, should develop a standardized risk assessment framework guidance document to evaluate the integration of artificial intelligence into nuclear stockpile stewardship and NC3 systems. 

This framework would enable systematic evaluation of risks, including confabulations, human-AI configuration, and information security, across modernization efforts. The framework could assess the extent to which an AI model is prone to confabulations, involving performance evaluations (or “benchmarking”) under a wide range of realistic conditions. While there are public measurements for confabulations, it is essential to evaluate AI systems on data relevant to the deployment circumstances, which could involve highly sensitive military information.  

Additionally, the framework could assess human-AI configuration with specific focus on risks from automation bias and the degree of human oversight. For these tests, it is important to put the AI systems in contact with human operators in situations that are as close to real deployment as possible, for example when operators are tired, distracted, or under pressure. 

Finally, the framework could include assessments of information security under extreme conditions. This should include simulating comprehensive adversarial attacks (or “red-teaming”) to understand how the AI system and its human operators behave when subject to a range of known attacks on AI systems.

NNSA should be included in this development due to their mission ownership of stockpile stewardship and nuclear safety, and leadership in advanced modeling and simulation capabilities. DARPA should be included due to its role as the cutting edge research and development agency, extensive experience in AI red-teaming, and understanding of the AI vulnerabilities landscape. STRATCOM must be included as the operational commander of NC3 systems, to ensure the framework accounts for real-word needs and escalation risks. OASD (ND-CBD) should be involved given the office’s responsibilities to oversee nuclear modernization and coordinate across the interagency. The OUSD (P) should be included to provide strategic oversight and ensure the risk assessment aligns with broader defense policy objectives and international commitments.

Recommendation 2. CDAO should implement the Risk Assessment Framework with STRATCOM

While NNSA, DARPA, OASD (ND-CBD) and STRATCOM can jointly create the risk assessment framework, CDAO and STRATCOM should serve as the implementation leads for utilizing the framework. Given that the CDAO is already responsible for AI assurance, testing and evaluation, and algorithmic oversight, they would be well-positioned to work with relevant stakeholders to support implementation of the technical assessment. STRATCOM would have the strongest understanding of operational contexts with which to apply the framework. NNSA and DARPA therefore could advise on technical underpinnings with regards to AI of the framework, while the CDAO would prioritize operational governance and compliance, ensuring that there are clear risk assessments completed and understood when considering integration of AI into nuclear-related defense systems.

Recommendation 3. DARPA and CDAO should join the Nuclear Weapons Council 

Given their roles in the creation and implementation of the AI risk assessment framework, stakeholders from both DARPA and the CDAO should be incorporated into the Nuclear Weapons Council (NWC), either as full members or attendees to a subcommittee. As the NWC is the interagency body the DOE and the DoD responsible for sustaining and modernizing the U.S. nuclear deterrent, the NWC is responsible for endorsing military requirements, approving trade-offs, and ensuring alignment between DoD delivery systems and NNSA weapons. 

As AI capabilities become increasingly embedded in nuclear weapons stewardship, NC3 systems, and broader force modernization, the NWC must be equipped to evaluate associated risks and technological implications. Currently, the NWC is composed of senior officials from the Department of Defense, the Joint Chiefs of Staff, and the Department of Energy, including the NNSA. While these entities bring deep domain expertise in nuclear policy, military operations, and weapons production, the Council lacks additional representation focused on AI.

DARPA’s inclusion would ensure that early-stage technology developments and red-teaming insights are considered upstream in decision-making. Likewise, CDAO’s presence would provide continuity in AI assurance, testing, and digital system governance across operational defense components. Their participation would enhance the Council’s ability to address new categories of risk, such as model confabulation, automation bias, and adversarial manipulation of AI systems, that are not traditionally covered by existing nuclear stakeholders. By incorporating DARPA and CDAO, the NWC would be better positioned to make informed decisions that reflect both traditional nuclear considerations and the rapidly evolving technological landscape that increasingly shapes them.

Conclusion 

While AI is likely to be integrated into components of the U.S. nuclear enterprise, without a standardized initial approach to assessing and managing AI-specific risk, including confabulations, automation bias, and novel cybersecurity threats, this integration could undermine an effective deterrent. A risk assessment framework coordinated by OASD (ND-CBD), with STRATCOM, NNSA and DARPA, and implemented with support of the CDAO, could provide a starting point for NWC decisions and assessments of the alignment between DoD delivery system needs, the NNSA stockpile, and NC3 systems.

This memo was written by an AI Safety Policy Entrepreneurship Fellow over the course of a six-month, part-time program that supports individuals in advancing their policy ideas into practice. You can read more policy memos and learn about Policy Entrepreneurship Fellows here.

Frequently Asked Questions

Does the NWC have the authority to create a new subcommittee including DARPA and the CDAO?

Yes, NWC subordinate organizations or subcommittees are not codified in Title 10 USC §179, so the NWC has the flexibility to create, merge, or abolish organizations and subcommittees as needed.

read more

Are there existing regulations that the United States has declared with respect to AI integration into NC3?

Section 1638 of the FY2025 National Defense Authorization Act established a Statement of Policy emphasizing that any use of AI in support of strategic deterrence should not compromise, “the principle of requiring positive human actions in execution of decisions by the President with respect to the employment of nuclear weapons.” However, as this memo describes, AI presents further challenges outside of solely keeping a human in the loop in terms of decision-making.

read more

Washington, D.C. – May 6, 2025 – The Federation of American Scientists (FAS) today released “Nuclear Notebook: Russian Nuclear Weapons 2025,” its authoritative annual survey of Russia’s nuclear weapons arsenal. The FAS Nuclear Notebook is considered the most reliable public source for information on global nuclear arsenals for all nine nuclear-armed states. FAS has played a critical role for almost 80 years to increase transparency and accountability over the world’s nuclear arsenals and to support policies that reduce the numbers and risks of their use.

This year’s report, published in the Bulletin of the Atomic Scientists and by Taylor & Francis and available in full here, discusses the following takeaways:

• Russia currently maintains nearly 5,460 nuclear warheads, with an estimated 1,718 deployed. This represents a slight decrease in total warheads from previous years but still positions Russia as the world’s largest nuclear power alongside the United States.
• Russia continues to modernize its nuclear triad, replacing Soviet-era weapons with newer types, but modernization of ICBMs and strategic bombers has been slow. The country’s efforts to develop the advanced Sarmat (RS-28 or SS-29) ICBM and the next-generation strategic bomber, PAK DA, have faced delays and setbacks.
• The submarine-based nuclear force continues its modernization with Borei-class submarines replacing older types. A portion of Russian ballistic missile submarines are at sea at any given time on strategic deterrent patrols. Significant nuclear warhead and missile storage upgrades are underway at the Pacific and Northern fleet bases.
• Russia continues modernizing and emphasizing its nonstrategic  nuclear forces. This includes land- and sea-based dual-capable missiles and tactical aircraft. Despite modernization of launchers, the number of warheads assigned to those launchers has remained relatively stable. Russia held several high-profile exercises with its nonstrategic forces in 2024, and the authors describe upgrades to a suspected nuclear storage depot in Belarus.
• Russia has maintained its policy of nuclear deterrence, emphasizing the strategic importance of its nuclear arsenal in its military doctrine. Updates to public policy documents describe a broader range of scenarios for potential use of nuclear weapons but it is unknown to what extent this is reflected in changes to military plans.
  

FAS Nuclear Experts and Previous Issues of Nuclear Notebook

The FAS Nuclear Notebook, co-authored by Hans M. Kristensen, Matt Korda, Eliana Johns, and Mackenzie Knight, is published bi-monthly in the Bulletin of the Atomic Scientists. The first Nuclear Notebook was published in 1987. FAS, formed in 1945 by the scientists who developed the first U.S. nuclear weapon, has worked since to increase nuclear transparency, reduce nuclear risks, and advocate for responsible reductions of nuclear arsenal and their role.

This latest issue on Russia’s nuclear weapons comes after the release of Nuclear Notebook: Chinese Nuclear Weapons 2025 and will be followed in June by Nuclear Notebook: French Nuclear Weapons 2025. More research available at FAS’s Nuclear Information Project.


The Federation of American Scientists’ work on nuclear transparency would not be possible without generous support from the Carnegie Corporation of New York, the Joseph Rowntree Charitable Trust, the Jubitz Family Foundation, the New-Land Foundation, Ploughshares, the Prospect Hill Foundation, and individual donors.

###

ABOUT FAS

The Federation of American Scientists (FAS) works to advance progress on a broad suite of contemporary issues where science, technology, and innovation policy can deliver transformative impact, and seeks to ensure that scientific and technical expertise have a seat at the policymaking table. Established in 1945 by scientists in response to the atomic bomb, FAS continues to bring scientific rigor and analysis to address contemporary challenges. More information about FAS work at fas.org.

In 2014, my colleagues  at the Monterey Institute for International Studies and I authored a study called The Trillion Dollar Nuclear Triad. This report laid out comprehensively, for the first time, that the recently launched US nuclear modernization program–including building new nuclear-armed submarines, bombers, and long-range missiles as well as new nuclear weapons–would likely cost over $1 trillion during its 30-year time scale. Until then, no one had even an estimate for how much these programs would cost over their lifetimes. At the time, my co-authors and I were criticized by advocates for the nuclear modernization program as alarmist and accused of actively seeking to inflate estimates to undermine political support for these programs. Overall, proponents argued that these programs were necessary and that the cost was a mere fraction of the Pentagon’s overall budget, historically low when compared to the nuclear modernization of the 1970s and 1980s. What was originally described merely as a bow wave of costs has become a tsunami.

On April 25, the Congressional Budget Office released its latest estimate that costs of maintaining the current nuclear arsenal and modernizing the entire program would cost nearly $1 trillion over the next 10 years alone.  As was once said, there are lies, damn lies, and statistics. This is not an argument about statistics but about absolute numbers. By any measure, $1 trillion is an enormous amount of money to spend in 10 years, and any other U.S. Government or private effort on this scale would receive intense scrutiny and attention–in fact, the Sentinel ICBM modernization most recently triggered a critical breach of the Nunn McCurdy act in 2024. Moreover, if the nuclear budget is going to average 10% from the estimated $1 trillion annual defense budget for the next decade, the nuclear program is now eating the defense budget alive.

For those who may doubt whether oversight is necessary, one only needs to look at one of the more visible elements of the nuclear modernization program–the effort to replace the Minuteman III intercontinental ballistic missile program. When we did our original estimate in 2014, there was no estimate for the cost of replacing the MMIII. We noted that past efforts to build comparable long-range missile systems were complex, expensive, and prone to large cost overruns. The example we cited was the “Evolved Expendable Launch Vehicle, which involved the purchase of 150 space launch vehicles based on existing technologies. Originally slated to cost $30 billion, program costs now exceed $70 billion.”

The sole recommendation of this 2014 report was that the USG and Department of Defense in particular needs to create a stand alone nuclear budget to understand the complex nature of these programs and their costs, and keep them on budget and under control. While we did not push the recommendation in the report, we strongly suggested overall that the program was unlikely to meet schedule and cost milestones because it was going to put the constrained U.S. defense industrial capacity under stress and it would be better to stagger these complex multi-decade and multi-billion dollar programs.Ten years later, this recommendation is still needed.

To push these programs through and dismiss alternatives, the military and defense contractors and defense hawks in Congress argued the sky was falling, that Russia and China were modernizing, that nuclear war would be more expensive, so there was no margin for error or delay to consider alternatives. Ironically, the mismanagement of the modernization program now threatens to produce the very delay they were warning against.

In all of these areas, our report was not only rebuffed by advocates for the nuclear modernization program, but our recommendations were ignored. And over time, the consequences of plowing ahead with these programs despite predicted problems have come to pass. The MMIII replacement program, now dubbed Sentinel, was originally estimated to cost $62 billion. That program is now slated to cost $126 billion in the next 10 years alone, and over $140 billion in total. Those costs can also be expected to keep rising due to ongoing delays in the program. This program bypassed the normal programmatic reviews and procurement milestones, and was rushed to a single source contract during President Trump’s first term, with predictably bad results. It now seems likely that the Air Force will have to extend the life of the MMIII missile through 2050–something supporters of the new ICBM have argued would be impossible in both 2021 and 2023.

There is still no comprehensive budget from DOD laying out their costs or predicting the full cost of implementing the current U.S. nuclear weapons modernization program. This is akin to agreeing to buying a house without knowing the sales price, the mortgage rate, or the monthly payment. Yes, you need to live somewhere, but in an era where critical U.S. programs for defense, science, climate, health and international programs and even nuclear weapons security and safety are being cut, it is nonsensical that these programs continue without adequate oversight and knowledge of their true costs.

Whether the U.S. can afford these programs is a reasonable question. Knowing how much these new systems will cost, it is appropriate to ask: if the Pentagon can not produce even a budget on costs, how can they be expected to oversee and deliver what their advocates claim are critical national security programs? The inadequate oversight, cost management and even basic program accountability will result, as we predicted in 2014, “disarmament by default”,  an unsustainable and dangerous way to manage U.S. national security.

The United States Air Force has forward deployed about one-third of its B-2 stealth bombers to Diego Garcia, or about half the B-2s considered fully operational at any given time. A Planet Labs satellite image taken earlier today shows six of the characteristic bombers on the apron alongside six refueling tankers.

A Planet Lab satellite image from earlier today shows six B-2 bombers at Diego Garcia in the Indian Ocean.

The current deployment of at least six B-2s to Diego Garcia is unusually large and exceeds the number of climate tents at the base designed to protect the sensitive surface of the bombers. The current deployment began a week ago.

US bombers regularly deploy on so-called Bomber Task Force missions to bases overseas or long-range exercise flights. That has included occasional B-2 bombers to Diego Garcia before (one in 2024 (hot pit refuel) and three in 2020), but the visits are rare and B-2s are designed to carry out global strike missions without having to deploy to forward locations. A B-2 strike against the Houthis in November 2024, for example, was flown directly from Whiteman AFB.

B-52 bombers have also deployed to Diego Garcia, including in 2024.

During 2024, U.S. bombers (B-1, B-2, and B-52) were deployed about 60 percent of the year to a theater or providing continental U.S. (CONUS) -to-CONUS flights in support of theaters or in support of U.S. Strategic Command and the Chairman of Joint Chiefs of Staff,” according to the U.S. Air Force.

During the past week the number of bombers visible at the base has fluctuated from none to two or four, indicating that the remaining bombers are either inside the climate tents or airborne on a training or perhaps bombing mission against Houthis forces. The U.S. Air Force operates a fleet of 20 B-2s. Of these, one is a test aircraft, and normally only 10-12 are considered fully operational with the rest undergoing repairs and routine maintenance. During the Spirit Vigilance exercise in April 2024, for example, the 509^th and 131^st bomber wings at Whiteman AFB in Missouri conducted a “mass fly-off” of all the operational B-2 bombers.

The B-2s deployed at Diego Garcia constitute about half of the operational B-2s, seen here in a “mass fly-off” exercise at Whiteman AFB in 2024.

That means about half of the entire operational B-2 force is currently forward deployed at Diego Garcia. The large deployment to Diego Garcia is part of evolving B-2 operations in the Western Pacific and Indian Ocean that also involve more deployments to Australia.

The B-2 bombers are dual-capable, which means they can deliver both conventional and nuclear weapons. Conventional weapons include precision guided bombs and cruise missiles as well as heavy earth penetrators. Nuclear weapons include the B61-12 guided bombs and the B61-11 nuclear earth-penetrator (the B61-7 has been replaced by the B61-12 and the B83-1 is no longer operational). There are no nuclear weapons on this Diego Garcia mission.

---

Background: United States Nuclear Weapons, 2025

Washington, D.C. – March 12, 2025 – The Federation of American Scientists (FAS) today released “Nuclear Notebook: China” – its authoritative annual survey of China’s nuclear weapons arsenal. The FAS Nuclear Notebook is considered the most reliable public source for information on global nuclear arsenals for all nine nuclear-armed states. FAS has played a critical role for almost 80 years to increase transparency and accountability over the world’s nuclear arsenals and to support policies that reduce the numbers and risks of their use.

This year’s report, published in the Bulletin of Atomic Scientists, shows the following nuclear trends:

• The total number of Chinese nuclear warheads is now estimated to include approximately 600 warheads. The vast majority of these are in storage and a small number—perhaps 24—are deployed. 

• China is NOT a nuclear “peer” of the United States, as some contend. China’s total number of approximately 600 warheads constitutes only a small portion of the United States’ estimated stockpile of 3,700 warheads.  While the United States has a fully established triad of strategic forces, China is still working to develop a nuclear triad; the submarine-based leg has significantly less capability, and the bomber leg is far less capable than the United States.
  
• China continues to see their land-based intercontinental ballistic missiles (ICBMs) as their most reliable and survivable nuclear force. China continues to prioritize its land-based force and increase its role in both number and capability. China has completed construction of its three new ICBM silo fields, one of which was publicly disclosed by FAS in 2021.  We estimate that around 30 silos have been loaded. China has also increased the number of road-mobile ICBM bases.
  
• China is continuing to develop its relatively small ballistic missile submarines (SSBN) force through improved missiles and a follow-on SSBN class. Production of the new “Type 096” SSBN has been delayed. We estimate that Chinese SSBNs are now conducting continuous patrols with nuclear weapons on board. However, Chinese SSBNs cannot target continental United States from their operating areas.
  
• Development of a viable nuclear bomber force is still in its early stages. Only one base has been established at Neixiang for the new H-6N bomber. The H-20 replacement bomber appears to be delayed.  

FAS Nuclear Experts and Previous Issues of Nuclear Notebook

The FAS Nuclear Notebook, co-authored by Hans M. Kristensen, Matt Korda, Eliana Johns, and Mackenzie Knight, is published bi-monthly in the Bulletin of the Atomic Scientists. The joint publication began in 1987. FAS, formed in 1945 by the scientists who developed the nuclear weapon, has worked since to increase nuclear transparency, reduce nuclear risks, and advocate for responsible reductions of nuclear arsenal and their role.

This latest issue on the United State’s nuclear weapons comes after the release of Nuclear Notebook: United States on America’s nuclear arsenal. More research available at FAS’s Nuclear Information Project.

###

ABOUT FAS

The Federation of American Scientists (FAS) works to advance progress on a broad suite of contemporary issues where science, technology, and innovation policy can deliver transformative impact, and seeks to ensure that scientific and technical expertise have a seat at the policymaking table. Established in 1945 by scientists in response to the atomic bomb, FAS continues to bring scientific rigor and analysis to address contemporary challenges. More information about FAS work at fas.org.

Washington, D.C. – March 7, 2025 – The Federation of American Scientists (FAS) is pleased to welcome Dr. Yong-Bee Lim as the new Associate Director of Global Risk. In this role, Dr. Lim will help develop, organize, and implement FAS’s growing contribution in the area of catastrophic risk prevention, including on core areas of nuclear weapons, AI and national security, space and other emerging technologies.  

“The role of informed, credible and engaging organizations in support of sound public policy is more important than ever” said Jon Wolfsthal, FAS Director of Global Risk. “Yong-Bee embodies what it means to be an effective policy entrepreneur and to make meaningful contributions to US and global security. We are really excited that he is now part of the FAS team.”

Dr. Lim is a recognized expert in biosecurity, emerging technologies, and converging risks through his former roles as Deputy Director of both the the Converging Risks Lab and the Janne E. Nolan Center at the Council on Strategic Risks, his research and leadership roles in academia, and through his work at key agencies (DoD, HHS/ASPR, and DoE) in the United States. He completed his Ph.D. in Biodefense from George Mason University’s Biodefense program, where he conducted critical work on understanding the safety, security, and cultural dimensions of the U.S.-based Do-It-Yourself Biology (DIYBio) community. His recent accolades include being in the inaugural fellowship class of the Editorial Fellows program at the Bulletin of the Atomic Scientists and his selection and involvement in the Emerging Leaders in Biosecurity Initiative hosted by the Johns Hopkins Center for Health Security. 

“As emerging capabilities change the very contours of safety, security, and innovation, FAS has positioned itself to both highlight the global opportunities we must seize and address the global risks we must mitigate,” Lim said. “Founded in 1945, FAS continues to display thought leadership and impact because it has not forgotten its core mission: to ensure that scientific and technical expertise continue to have a seat at the policymaking table. I am honored to be part of an organization with a legacy and mission like FAS.”

ABOUT FAS

The Federation of American Scientists (FAS) works to advance progress on a broad suite of issues where science, technology, and innovation policy can deliver transformative impact, and seeks to ensure that scientific and technical expertise have a seat at the policymaking table. Established in 1945 by scientists in response to the atomic bomb, FAS continues to bring scientific rigor and analysis to address contemporary challenges. More information about FAS work at fas.org and Global Risk, here.

Between 2006 and 2015, the United Kingdom repeatedly and publicly announced its intentions to decrease the size of its nuclear weapons stockpile, most recently committing to at most 180 weapons by the mid-2020s. As the years went by, non-government policy experts and nuke watchers assumed that the UK Government was making good on its word and that the UK nuclear arsenal would continue to gradually reduce. In reality, the United Kingdom is thought to have maintained a nuclear stockpile of approximately 225 weapons, and, in a surprise move in 2021, the United Kingdom declared that it would extend the ceiling of its “overall nuclear weapon stockpile” to no more than 260 weapons. This constituted an abrupt about-face from its previous commitments and trajectory. 

However, the infrastructure underpinning the sustainability of the United Kingdom’s nuclear weapons and their modernization has experienced significant budgetary and scheduling challenges. Furthermore, the UK has dramatically reduced the public transparency of its nuclear forces, making it increasingly difficult to understand and debate the true scope of these challenges.

The UK’s Nuclear Warhead Modernization

Announced in 2005, the Atomic Weapons Establishment’s (AWE) Nuclear Warhead Capability Sustainment Programme (NWCSP) was an initiative to deliver infrastructure and technology to sustain the United Kingdom’s current stockpile and underpin its warhead replacement program. Each of these main infrastructure and technology projects is named after a constellation: Project Orion for a high-power research laser that began operations in 2013, Project Leo for a small parts manufacturing facility, Project Pegasus for the manufacturing of uranium components effort, and so on. Several of these projects under the NWCSP related to techniques used for nuclear weapons development in place of explosive testing. Part of the NWCSP mission also involved refurbishing the United Kingdom’s current warheads for integration with the U.S.-supplied, upgraded Mk4A aeroshell, which was completed in 2023. The NWCSP was scheduled to run from April 2008 until April 2025 and was removed from the MOD Government Major Projects Portfolio data starting in 2022, indicating it could have been downsized from a large-scale development initiative.

In February 2020, the UK defence secretary announced a new warhead program—the A21/Mk7/ Astraea. This new warhead is currently in the concept stage but is planned to eventually replace the Mk4A/Holbrook beginning in the late 2030s. As was the case with the previous version, the A21/Mk7 Astraea design and production will have a “very close connection” to the future US W93/Mk7.

In 2023, the MOD confirmed that £127 million had been spent on the A21 Astraea warhead replacement program as of March 2022. The total cost of the replacement program has not been released, given it is still in the early stages. However, even if cumulative costs are released, individual costs associated with the warhead modernization program will be challenging to identify due to changes in UK budgetary reporting practices. In 2023, nuclear-related programs and expenditures—including the AWE and the NWCSP—were compiled into one heading under Defence Nuclear Enterprise (DNE), which appears as a single line item in the departmental estimates. This makes it impossible to see the direct in-service costs associated with the individual programs. DNE funding was also “ringfenced” within the MOD budget to protect it against spending cuts. 

The UK’s declaration that its stockpile ceiling will be raised to 260 raises several questions. In the early- to mid-2000s, the United Kingdom possessed a stockpile of roughly 280 warheads, but in 2010 announced it would reduce this level to “no more than 225 warheads.” To be able to increase this level to up to 260 warheads, as announced by the UK government in 2021, it seems some of the retired warheads—or components from them—must have been retained in some form. The United Kingdom appears to use the term “stockpile” to describe operational, deployed, and retired weapons. The timeline for potentially increasing the stockpile to up to 260 is not known, but if it is a relatively short timeline, it would seem to require reconstituting some retired warheads. If it is a longer timeline, it could potentially involve increasing the number of A21 Astraea warheads in the future. Of the “no more than 225” warhead stockpile level, the United Kingdom has previously explained that 120 warheads would be operationally available, of which 40 would be deployed on the single ballistic missile submarine (SSBN) on patrol at all times. The reason for increasing the stockpile to up to 260 warheads appears to be derived from an interest in increasing the number of “operationally available” warheads to be able to deploy a full warhead load on the SSBN fleet—something the recent warhead level did not allow.

Critical Infrastructure 

Any warhead design, manufacturing, and testing occurs at the AWE site at Aldermaston while the warheads are assembled, maintained, and decommissioned at AWE Burghfield. These two sites are critical to maintaining the United Kingdom’s existing stockpile and will play a significant role in the new warhead replacement program. 

Upgrades at AWE Burghfield

Project MENSA—one of the infrastructure programs under the NWCSP—aims to consolidate existing nuclear warhead assembly and disassembly operations into a single building located in the center of the AWE Burghfield complex called the Main Processing Facility (MPF). MENSA will replace the existing Gravel Gertie bunkers used for warhead assembly and disassembly on the eastern part of the campus, which are designed to collapse inward in the event of an explosion. Other new infrastructure for this project includes a support building and 16 lightning protector towers to accompany the MPF, as well as an associated plant building, gatehouses, vehicle inspection bays, substation buildings, security fences, access roads, and Sustainable Drainage System (SuDS) infrastructure.

Project MENSA’s completion has been delayed by more than seven years, and its expected cost is £1.36 billion over its original total budget of £0.8 billion. Construction progress at the site can be observed through satellite imagery and should be nearing completion of construction according to the MoD’s 2024 Major Project Portfolio data. 

Upgrades at AWE Aldermaston

AWE Aldermaston, where warhead design, development, manufacturing, assembly, and testing occur, is also undergoing significant upgrades and revitalization. In 2024, AWE announced two new infrastructure programs—the Future Infrastructure Programme (FIP) for general infrastructure and the Future Materials Campus (FMC) for nuclear material manufacturing and storage—to consolidate existing programs and invest in new ones to increase capacity for maintaining, manufacturing, and storing nuclear weapons. The procurement process for these multi-year, multi-billion pound projects began on April 22 and December 12, 2024, respectively. The FMC, in particular, will be a collection of facilities, including nuclear science and technology centers and laboratories, to be built at AWE Aldermaston. This program will replace two major projects that initially fell under the NWCSP—Project Pegasus and Project Aurora. 

Project Pegasus was described as a new enriched uranium storage and manufacturing facility at AWE Aldermaston that would replace the existing enriched uranium handling facility located at the A45 building. Work began in 2003, and the original projected service date was 2016. The approved cost was originally £634 million before it skyrocketed to £1.7 billion. After a six-year delay and a three-year pause, construction of the new storage facility began, and the manufacturing facility was scheduled to be finished by 2030. The severe delays were mainly due to challenges with the supply chain environment and an “overly complex technical solution” that resulted in additional construction and safety costs and a “reassessment” of the project design and requirements. 

The UK government was also in the early design phase of Project Aurora, an infrastructure project for a new plutonium manufacturing facility that would replace the current A90 building at AWE Aldermaston. This program was experiencing delays due to resourcing problems, supply chain shortages, and high-skilled workforce challenges. Project Aurora was removed from the NWCSP and added as an independent program to the MOD’s Major Projects Portfolio in 2022. Projects Aurora and Pegasus were both removed from the 2024 version of the MOD’s Major Projects Portfolio database, likely due to their absorption into the new FMC program.

The other central element of NWCSP is the delivery of a new hydrodynamics facility. In 2010, the United Kingdom and France signed the Teutates Treaty, which allowed for cooperation on warhead physics research between the two nations. From this agreement, the EPURE radiographic facility was built at Valduc in France, and a joint UK-France Technology Development Centre was established at Aldermaston. These facilities will support hydrodynamic research that will allow the study of the effects of aging and manufacturing processes on nuclear warheads without nuclear explosive testing. Because of the Teutates program, the UK’s original plans for a ‘Project Hydrus’ hydrodynamics facility were canceled. 

Outstanding Challenges 

The facilities described above are deemed crucial to the United Kingdom’s effort to modernize and refurbish its nuclear deterrent. However, the significant delays and cost overruns have elicited criticism from the public and concern from government authorities over the years. In August 2024, the House of Commons reported a spending deficit on nuclear programs across the Defence Equipment Plan. Over the next ten years (2023 to 2033), costs are forecasted at £117.8 billion, of which only £109.8 billion had been budgeted for when the report was released. 

Alongside these warhead development and infrastructure programs, the United Kingdom is also replacing its four SSBNs with the new Dreadnaught class, the first of which is expected to enter service in the early 2030s. This modernization has also been plagued by cost increases and threatens to be delayed, given maintenance issues with the existing fleet.

Moreover, the recent decrease in UK government transparency regarding the status of its nuclear arsenal and modernization program reflects a worrisome global trend. The United Kingdom has not publicly declared the approximate size of its stockpile since 2010, following the Obama Administration’s decision to release its stockpile number. In 2021, the MOD referred to the 2010 announcement again but did not explicitly say what the stockpile number was. Moreover, it said it would “no longer give public figures for our operational stockpile, deployed warhead or deployed missile numbers.”

Additionally, the MOD said in 2023 that it was withholding information on planned in-service dates for many of the above-referenced projects for “reasons of national security” and did not release this information in its 2022, 2023, and 2024 Major Projects Portfolio data. Finally, the MOD has published an annual update to Parliament since 2011 on the progress of nuclear weapons upgrades, but there has been no report published for 2023 or 2024.

While some of this backsliding is likely rooted in concerns about public perception of persistent programmatic delays, the overall picture raises concerns about a pattern of declining transparency and reduced public ability to monitor and hold the UK government accountable for its nuclear weapons program. 

In the spring of 2022, researchers at the Federation of American Scientists began reading newly released U.S. Defense Department budget documents to look for updates concerning the Pentagon’s priorities for the next fiscal year. As the researchers poured over hundreds of pages, two words suddenly captured their attention: the Biden administration’s Fiscal Year (FY) 2023 budget request had added “the UK” to a list of countries receiving upgrades to their “special weapons” storage sites under a 13-year NATO investment program. The term “special weapons” is often used by the U.S. government when referring to nuclear weapons. However, the United States has not deployed nuclear weapons in the United Kingdom for nearly two decades. Those two words sparked dozens of questions, years of continued research, and a new local movement of protests against the return of a potential nuclear mission to RAF Lakenheath.

This new report provides an account of the nuclear history of RAF Lakenheath and the role it played in the US nuclear mission until nuclear weapons were withdrawn in 2008. The report then explains the mounting evidence from three years of collection of documentation and observations that show the United States Air Force is re-establishing its nuclear mission on UK soil for the first time in nearly two decades.

As of February 2025, there are no known public indications that nuclear weapons have been deployed to RAF Lakenheath – we assess that the return of the nuclear mission is intended primarily as a backup rather than to deploy weapons now. However, if this were to happen, it would break with decades of policy and planning and reverse the southern focus of the European nuclear deployment that emerged after the end of the Cold War. Even without weapons present, the addition of a large nuclear air base in northern Europe is a significant new development that would have been inconceivable just a decade-and-a-half ago.

Read the full report here

---

The authors would like to thank Matt Korda and Kate Kohn for their invaluable contributions to this report, as well as the Joseph Rowntree Charitable Trust for their support.

President Trump is moving quickly to push U.S. allies to invest even more in their own defense. NATO allies have already committed to spend 3% of their GDP on defense, yet the U.S. is now calling for them to spend at least 5%. It is likely that U.S. allies in East Asia will soon face similar calls to do more. Greater investments in conventional capabilities make a lot of sense. However, there are some U.S. policy experts, officials and academics calling for more U.S. allies to go nuclear to reduce U.S. defense requirements. These calls are dangerously misguided and ignore the threat any proliferation – including by U.S. allies – poses to American security interests. They must be rejected wholesale by the Trump Administration.  

One of the most enduring successes of U.S. national security policy has been its effort to limit the number of states with nuclear weapons. Predictions that dozens of countries might possess nuclear weapons did not materialize because of concerted U.S. actions. The risks include the reality that U.S. allies can and often do experience internal instability or even regime collapse, that any state with nuclear weapons creates a risk that those materials or knowhow can be stolen or diverted, that any state with nuclear weapon in a crisis might actually use those weapons, and lastly the reality that states with their nuclear weapons are less susceptible open to U.S. influence. There may be reasons why a state may want to go nuclear from their own perspective but there are few if any lasting benefits to American security that comes from proliferation to friends and allies.

Nine countries currently have nuclear weapons, but perhaps 40 additional states are technically advanced enough to build nuclear weapons if they chose to do so. Many of these states are U.S. allies or partners, including in Europe as well as Japan, South Korea, and even the island of Taiwan. That these states never went nuclear (although some tried) is due to a combination of factors, including the credibility of U.S. defense commitments to their security, the pressure America brought to bear when these states indicated a potential interest in building independent nuclear arsenals, and the recognition that if the world was serious about getting rid of all nuclear weapons then their spread was a step in the wrong direction.

The re-election of Donald Trump has understandably spooked many U.S. allies, renewing doubts that America will come to their aid. The growth of China’s military and economic power relative to the United States is adding to these concerns. More allies are asking now, just as they did during the Cold War if America would really risk Boston to protect Berlin, or Seattle to protect Seoul. As this question festers and as America’s relative power over China and other states ebbs, the lure to encourage U.S. friends to develop nuclear weapons of their own to deter or defeat an attack will grow. After all, the theory goes, why should the United States worry if its friends go nuclear?

In the real world, however, the spread of nuclear weapons anywhere complicates and undermines U.S. security. One reason is states are not always stable. In the 1970s, the U.S. supported its Treaty partner Iran acquiring nuclear reactors and advanced technology but in 1979, that regime was overthrown by the Islamic Revolution. Pakistan went nuclear when the U.S. needed its help fighting the Soviets in Afghanistan, and has faced wave after wave of instability and crisis. And South Korea is a more recent challenge. For the last few decades, South Korea was considered a stable and vibrant democracy – even hosting a Summit for Democracy last year. Under President Yoon, South Korea has voiced increasing interest in an independent nuclear arsenal. And just last year, a former Trump official, Elbridge Colby, expected to serve in a senior policy role at the Pentagon publicly encouraged South Korea to build their own nuclear weapons to deter North Korea and enable the U.S. to focus more on China. The situation in South Korea, with an impeached President and no clear sense of who controls the country’s military, would be a lot more dangerous if Seoul had nuclear weapons. 

This is not just an issue for newer nuclear weapons states. Prior to the Soviet Union’s collapse in 1991, a coup created confusion for days over exactly who had the ability to control Soviet nuclear weapons.  Following the USSR’s demise, nuclear weapons and materials remained at risk of theft and diversion for years and required massive U.S. efforts and investments to prevent their loss. And even the United States is not immune from these risks. The 2021 insurrection raised nuclear risk to the point that the Speaker of the House had to publicly ask the Chairman of the Joint Chiefs about the risk that President Trump might use nuclear weapons in a gambit to remain in power, and Chairman Milley took extraordinary steps to insert himself into the nuclear chain of command to preempt that risk. Any nuclear arsenal anywhere is a potential danger if political circumstances change.

And states with nuclear weapons create a nuclear risk if nuclear technology, materials and knowhow are stolen or diverted. Five of today’s nuclear weapon states – America, Russia, China, France, and Pakistan – have either knowingly or unwittingly helped other states go nuclear.  Even if theft or transfer were not an issue, when new states have gone nuclear in the past, others have followed. America’s nuclear success led the Soviet Union to build them as well. This in turn led the UK and France to follow suit. These four nuclear weapon programs fueled China’s desire to join the club. Beijing having the bomb drove India to do the same, which then led Pakistan to follow suit.  

And any nuclear state might decide one day to use those weapons. Every nuclear leader must get every nuclear decision right, every time or boom. The history of U.S. and Soviet nuclear deterrence is marked as much by nuclear misunderstandings and potential accidents as by stable deterrence. India and Pakistan have the same problem. It is reasonable to assume new nuclear states with nuclear weapons would encounter many of the same risks.

And finally, from a very direct Americentric point of view, each state that acquires their own nuclear weapons lessens the ability of the United States to influence, control or dictate security outcomes in that state and region. While not the message U.S. diplomats use openly when trying to work diplomatically to stop proliferation, the issue of influence is as relevant to U.S. allies as adversaries. To the extent that the U.S. security is enhanced by being able to heavily influence how states around the world act, then enabling the spread of nuclear weapons undermines that ability.

It is and will continue to be tempting for the next Administration to find rapid and easy solutions to long-standing security challenges. Empowering U.S. allies to do more so Washington can do and spend less, or focus more effectively on fewer challenges is an understandable policy outcome. But enabling, or looking the other way at the spread of nuclear weapons is not in America’s interests anymore today than it was in the 20th century.