Optimal Deterrence

Executive Summary

The United States faces growing dangers of nuclear escalation, a new arms race, and proliferation. These risks stem, in part, from its strategy of using its nuclear forces to target opponents’ nuclear forces. Such “counterforce” targeting is justified primarily as a way to limit the damage the United States would suffer in a nuclear war. However, adversaries’ nuclear forces are too difficult to destroy for this strategy to yield meaningful benefits, while its risks are high.

A new arms race. China is building up its nuclear forces. Russia may do so too. To meet counterforce targeting requirements, potentially against both adversaries simultaneously, the United States will need a larger nuclear force. Such a build-up, however, would motivate China and Russia to further increase their force requirements, thus stimulating an expensive, tension-generating, and futile three-way arms race in which the United States is poorly positioned to compete given the limitations of its defense industrial base.

Escalation. Counterforce targeting increases the likelihood that, in a conflict, an adversary could escalate the conflict because it feared its nuclear forces were vulnerable. It could, for example, engage in limited nuclear use or take actions, such as dispersing its nuclear forces, that risked catalyzing further escalation.

Proliferation. U.S. allies fear being abandoned by the United States, and some have begun to openly contemplate acquiring nuclear weapons. However, if Washington decides to rebuild its alliances, it may try to “assure” allies by augmenting its nuclear capabilities in a way that risks accelerating a new arms race.

As part of an improved strategy of “optimal deterrence,” the United States should

• cease, and declare it has ceased, the targeting of adversaries’ nuclear forces, command control systems, and leadership, and instead focus exclusively on conventional military forces and war supporting industry, which it already targets;
• weigh the pros and cons, in war planning, of not conducting conventional or nuclear operations that could convince an adversary that the United States was planning to attack its nuclear forces; and
• continue critical ongoing nuclear modernization programs.

The United States should also rebuild its relationships with allies, both because alliances bolster U.S. security and because abandoning allies would risk sparking proliferation. That said, the United States should not allow allies to practically dictate U.S. force requirements. To this end, Washington should

• acquire the capabilities—including nonnuclear capabilities—required to deter allies’ adversaries;
• apply a clear-eyed cost-benefit test to potential assurance measures that are unnecessary for deterrence; and
• make clear to allies that if they acquire nuclear weapons, they will lose U.S. extended deterrence guarantees and will be subject to the sanctions mandated by U.S. domestic law.

Introduction

For eighty years, the United States has sought to maximize the deterrent value of its nuclear arsenal while minimizing the risks of escalation, arms racing, and proliferation. Today, those risks are growing—indeed, they have been for some time—and the trend is set to continue. The United States is not the sole cause of that development, but neither is it blameless.

Newly emerging dangers of escalation and arms racing have their origins, in part, in how the United States plans to use nuclear weapons, if it has to. Among the small subset of Americans who have ever contemplated the arcane question of nuclear targeting, most probably assume that U.S. nuclear weapons would fall on an enemy’s population. In fact, the United States insists that it does not intentionally target civilians but would instead use nuclear weapons to further military objectives, such as limiting the damage the United States would suffer in a nuclear war.^[1] To that end, targets for U.S. nuclear weapons include adversaries’ intercontinental ballistic missiles (ICBMs); ballistic missile submarines (SSBNs); heavy bombers; nuclear bases; and nuclear command, control, communications, and intelligence (C3I) systems. Enemy nuclear forces are not the only targets, but they are the most numerous and thus largely determine how many nuclear weapons the United States needs.

Such “counterforce” targeting has led to an emerging bipartisan consensus that the United States’ nuclear force is too small. The reasoning of those who advocate for more U.S. nuclear weapons is straightforward: because “our” nuclear weapons target “theirs,” the more “they” have, the more “we” need.^[2] That logic is likely to gain traction as China’s nuclear buildup continues and even more so if Russia starts a buildup of its own.

The belief that the United States faces a deteriorating security environment, including growing and improving nuclear weapon capabilities on the part of its adversaries, is well founded; the proposed remedy of acquiring more nuclear weapons is not. Deploying more warheads or delivery systems would likely catalyze a costly and futile arms race. Not only would such an arms race be expensive, but it would also exacerbate tensions with China and Russia, thus increasing the likelihood of a conflict with either or both. Moreover, the United States would start a new arms race at a significant disadvantage: because the U.S. government needs to modernize its nuclear forces with an already overstretched industrial base, it is ill-equipped to compete. Even more seriously, counterforce targeting exacerbates the risk of escalation—from a conventional conflict to a limited nuclear war to an all-out nuclear war—by pressuring an adversary to use its nuclear weapons while it still could or to take provocative or dangerous steps to ensure the survivability of its nuclear weapons.

The long-standing U.S. goal of deterring threats to its allies—extended deterrence—adds additional challenges. Especially since the reelection of President Donald Trump, U.S. allies have feared being abandoned by the United States, and some have begun to openly contemplate acquiring nuclear weapons. Proliferation, however, even among U.S. allies, would be dangerous. It would raise the likelihood of nuclear war. Moreover, by exacerbating tensions between would-be nuclear-armed states and their rivals—which could ultimately lead rivals to attack a proliferator’s nuclear program preventively—it could cause conventional conflict, too. But if Washington decides to rebuild its alliances—as it should, and not only to further the objective of nonproliferation—there is a real risk its approach will focus on augmenting U.S. nuclear capabilities in a way that risks further accelerating a new arms race.

All that said, the future direction of U.S. policy is currently unclear. Many senior Trump administration officials appear sympathetic to calls for a U.S. nuclear buildup, and at least some value U.S. alliances. The president himself, by contrast, clearly doubts the value of alliances and seems instinctively skeptical about the case for more nuclear weapons, as recent remarks of his exemplify:

We’re going to have [China and Russia] spend a lot less money [on defense]. We’re going to spend a lot less money. And I know they’re going to do it. They agreed to it. We were talking about de-nuking, de-nuclearize, de-nuking. And President Putin and I agreed that we were going to do it in a very big way. There’s no reason for us to be building brand-new nuclear weapons.^[3]

Meanwhile, uncertainty about U.S. nuclear policy is compounded by questions about the affordability and feasibility of a rapid buildup. The United States’ nuclear weapon program faces deep fiscal challenges. To give one illustration, the cost of developing a new ICBM is now estimated to be 81 percent over the original budget.^[4] As a result, spending will need to increase just to execute the modernization program necessary to keep the U.S. nuclear force at its current size.^[5] Expanding the force will require more money still.

Against this background, the United States should optimize its nuclear strategy—to maximize the  deterrence benefits of its nuclear arsenal while reducing the risks of arms racing, escalation, and proliferation as well as the costs of modernization.

The starting point for that optimal strategy is the realization the United States’ nuclear arsenal already meets the two criteria for maximum efficacy. First, it is sufficiently survivable and reliable to inflict unacceptable damage on any adversary or combination of adversaries under any plausible circumstance. If Washington makes sensible investments in modernization, it will retain that ability without a buildup.

Second, with the three low-yield nuclear capabilities that the United States already has (all recently modernized or due to be soon), it can credibly conduct limited nuclear strikes, which would be particularly important in a regional war fought to protect an ally either in response to limited nuclear use by the adversary or because the United States viewed such use as the only way to stave off defeat.^[6] 

Even though the U.S. arsenal meets both criteria, nuclear deterrence could still fail, but neither counterforce targeting nor the larger nuclear arsenal that strategy necessitates would lessen the likelihood of such a failure or help the United States better manage its consequences. The hackneyed adage that a nuclear war cannot be won contains more than a kernel of truth.

Given the high costs and limited benefits of counterforce targeting, the United States should abandon the policy. Instead, the United States should aim its nuclear forces exclusively at adversaries’ conventional military forces and war-supporting industry (which the United States almost certainly already targets). That change would mitigate the risks of escalation and arms racing without weakening deterrence. Indeed, having rejected counterforce targeting, Washington could and should categorically foreswear a nuclear buildup, daring its adversaries to expend vast resources competing against no one in an imaginary competition with no finish line.

To curtail the danger of proliferation among U.S. allies, the United States should first decide it values alliances and try to rebuild them. But it cannot afford to go back to allowing allies’ appetite for assurance—efforts to convince allies that Washington has their backs—to fuel a U.S. nuclear buildup. Any benefits to assurance from the United States’ fielding more or different capabilities are likely to be too transient to be worthwhile. Instead, the United States should work with allies to ensure their adversaries are deterred, while taking a harder line against proliferation: it should make clear to its allies that acquiring nuclear weapons would cost them U.S. security guarantees.^[7]

The Security Environment: Bad and Getting Worse

The case for a larger and more diversified U.S. nuclear force starts with a largely correct description of an already dangerous security environment that is deteriorating further. Three features of that environment are particularly relevant.^[8]

First, while U.S. relations with China and Russia ebb and flow, there appears to be a generally rising risk of a conventional conflict between the United States and China over Taiwan and between the United States and Russia over U.S. allies, particularly the Baltic states. Conflict with either adversary could escalate into a nuclear war. (For the foreseeable future, North Korea’s nuclear arsenal will be far less consequential for the future development of the United States’ nuclear force than China’s or Russia’s will be.)

A second factor is the trajectories of the Chinese and Russian nuclear programs. The U.S. Department of Defense estimates that the number of Chinese “operational nuclear warheads” increased from the low two hundreds in 2020 to more than six hundred in 2024.^[9] It further assesses that China will have “over 1,000” such warheads by 2030 and will continue to grow its stockpile thereafter.^[10] As a point of comparison, the New Strategic Arms Reduction Treaty (New START) limits the U.S. nuclear force to 1,550 warheads deployed on SSBNs, ICBMs, and heavy bombers, out of  total arsenal of 3,748 deployed and nondeployed warheads in 2023.^[11]

China is also increasing the number, diversity, and sophistication of both its long-range and regional delivery systems, while also changing the way it operates its nuclear force.^[12] Historically, China has kept its nuclear warheads and delivery systems separate in peacetime, thus denying itself the ability to launch at short notice. Today, however, the Pentagon assesses that China keeps a portion of its nuclear force on “high duty alert,” with warheads mated to missiles, and it expects that practice to become more common in the future.^[13]

Russia is also improving its nuclear arsenal qualitatively, including through its development of myriad new long-range delivery systems. However, assessments published by the U.S. Departments of Defense and State suggest that, despite its violation of New START—by unilaterally terminating that treaty’s verification regime—Russia is not currently building up its nuclear force.^[14] Whether its abstinence will continue, especially after New START expires in February 2026, is far from clear.^[15] Indeed, in response to Trump’s goal of developing an American “Golden Dome”—a missile defense system capable of intercepting “any foreign aerial attack on the Homeland”—the Kremlin has already threatened to expand its nuclear force.^[16]

Arms control is unlikely to halt those dynamics. A U.S.-Russian agreement to replace New START may be within reach (though it is unlikely to be legally binding, permit onsite inspections, or limit Russia’s stockpile of nonstrategic nuclear weapons).^[17] If the United States and Russia entered into a bilateral arms-limitation agreement, however, the United States would still worry about China’s buildup; indeed, it would likely worry more given the limits on its own nuclear force. Yet the prospects for negotiating with China, let alone concluding a trilateral agreement, are bleak. Beijing has consistently rejected arms control in the past and, given its ambitious nuclear modernization program, is likely even less interested now. Moreover, even if Beijing were willing to negotiate, talks would likely be unproductive because Washington has not started to consider what concessions it would be prepared to make. In fact, a possible quid pro quo—limits on missile defenses, perhaps—would almost certainly be unacceptable to the United States, and even stringent and binding limits on missile defense would not necessarily induce China to agree to limits on offensive arms.

The third shift in the security environment is the growing closeness of China and Russia, including their proclaimed “no limits” partnership.^[18] That relationship has a nuclear dimension; Russia is assisting China with early-warning technology and providing nuclear fuel for reactors that the United States believes will be used to support China’s nuclear weapon program.^[19] Looking forward, U.S. military planners fear that, over time, the China-Russia relationship could grow into a military alliance and that, in the worst case, the United States would be forced to fight a two-front nuclear war. Even if such an alliance does not materialize, policymakers worry about the possibility of opportunistic aggression by one U.S. rival if the United States was fighting a war against the other.^[20] The likelihood of either a China-Russia military alliance or of opportunistic aggression is certainly debatable. However, over the four or more decades the next generation of U.S. nuclear weapons are likely to be in service, those possibilities are not so remote that military planners can safely ignore them.

The Costs of Counterforce: Arms Racing

The United States publicly professes that “a nuclear war cannot be won and must never be fought,” yet unclassified descriptions of its nuclear war planning suggest that, if the United States fights a nuclear war, it has no intention of losing.^[21] The Biden administration stated that it “would seek to end [a nuclear war] at the lowest level of damage possible on the best achievable terms for the United States and its allies and partners” and sought “flexible nuclear capabilities to achieve our objectives.”^[22] To that end, it promised to “maintain counterforce capabilities to reduce potential adversaries’ ability to employ nuclear weapons against the United States and its allies and partners.”^[23] The first Trump administration set out a similar policy but went even further in articulating the objective of “preventing further nuclear employment” by U.S. adversaries if the United States or a U.S. ally or partner were to face “a large-scale nuclear attack.”^[24]

In short, Washington wants the capabilities to attack opponents’ unused nuclear forces, their C3I systems, and perhaps their leaders. In launching such attacks, its objective would be to limit or even eliminate the adversary’s ability to inflict damage on the United States, its allies, and its partners. Supporters of the policy hope that such a damage-limitation capability makes it credible for the United States to risk nuclear war on behalf of an ally and thus extend deterrence.

Opponents’ nuclear forces are not the only targets for U.S. nuclear weapons. The United States almost certainly also aims to threaten adversaries’ conventional military forces, their war-supporting industry, and perhaps the internal security forces that keeps their leaders in power.^[25]That said, enemy nuclear weapons are more numerous and, accordingly, are the primary driver of the overall size of the United States’ nuclear force.

Given the worsening security environment, the United States’ current nuclear force is—as a matter of “simple logic and arithmetic”—too small to meet counterforce targeting requirements.^[26] Historically, it could field enough nuclear weapons to execute counterforce strikes on Russia while treating China as “a lesser included case.”^[27] Now that it worries about fighting a nuclear war against both at once, it “must,” in the words of Franklin Miller, an architect of U.S. nuclear strategy, “be able to threaten, separately and in combination, both Russia’s and China’s key assets,” including their nuclear forces.^[28] In practice, meeting that objective requires the United States to deploy additional warheads so its nuclear force becomes larger than either China’s or Russia’s.

The logic of counterforce targeting is merciless, however. If China continues to build up its nuclear force, as it appears set to do, then the United States will have to deploy more nuclear warheads to cover those additional targets. If Russia augments its nuclear forces, the United States will require yet more warheads.

For their part, China and Russia will worry that the U.S. buildup is compromising the survivability of their nuclear forces (after all, that is its purpose). Moreover, they will not accept the U.S. claim that it needs a larger force to deter joint action; instead, they will worry that the United States seeks superiority over each of them individually. In response, they will likely further increase their force requirements, which will prompt Washington to do the same.

The United States pursuit of Golden Dome is likely to exacerbate the pressures to build up. Beijing and Moscow both worry missile defenses could enable the United States to mop up any of their missiles that survived U.S. counterforce strikes. Their desire to undermine such defenses is likely to result in their fielding even larger forces.^[29] 

In short, the United States’ desire to credibly threaten China’s and Russia’s nuclear forces through counterforce operations backed up by missile defenses could well generate an arms race, in which the United States would waste enormous sums of money failing to achieve its objective.^[30] Because of long acquisition timelines and the fact that procurement plans are revised periodically and not continuously, that competition would play out in fits and starts over many years, but it would be an arms race nonetheless.

A three-way arms control agreement to head off such an arms race is hardly feasible. Historically, nuclear arms control was based on the principle of parity: agreements imposed roughly or exactly equal limits on Washington and Moscow. But that concept cannot be easily applied when three parties are involved. So long as the United States pursues counterforce targeting, it needs enough nuclear weapons to threaten China’s and Russia’s nuclear forces simultaneously and thus cannot accept an arms control treaty that limits it to the same number of nuclear weapons that China and Russia individually deploy. Neither China nor Russia, for their part, can accept an agreement that allows the United States to have more.^[31] (China has not historically felt the need for a large nuclear force, but that is quite different from signing onto an arms control agreement that would formally enshrine its inferiority.) In short, even if the United States, China, and Russia all supported arms control, it is difficult to see what numerical formula they could all accept.

Even under the best of circumstances, nuclear arms races are expensive, tense, and futile. For Washington, however, circumstances are far from the best. With an overstretched defense industrial base, the United States is badly positioned to compete. If it nonetheless competes, it will have to shortchange other defense priorities to fund nuclear programs whose costs are already ballooning.

The Trump administration has various options at its disposal for deploying more nuclear warheads in the short term.^[32] It could upload reserve warheads—that is, place them on existing delivery systems that can be loaded with more warheads than they currently carry. Additionally, under New START, the United States converted some launch tubes on SSBNs and some B-52H bombers so they could not be loaded with nuclear weapons. Reversing those conversions is another option. In total, the United States could deploy approximately one thousand additional warheads through those measures (the exact number is not publicly known).

Further significant increases would likely have to wait until at least the early 2040s. Many components of the United States’ nuclear force, including SSBNs and ICBMs, are now decades old and in genuine need of modernization. But following various delays, the last of the United States’ new Columbia-class SSBNs is now due to be deployed in about 2040, while the production of new Sentinel ICBMs may not be complete until 2050.^[33]

Those delays, particularly in ICBM modernization, could hinder the United States’ ability to compete in an arms race. In 2020, when the contract for the Sentinel program was awarded, the U.S. Air Force intended to keep Minuteman III ICBMs operational until 2036, when the last Sentinel ICBM was due to be deployed. Now, however, it needs to keep Minuteman III ICBMs in service for well over a decade longer. If the Air Force fails to do so, the size of the U.S. ICBM force will shrink before it recovers to its current level. By contrast, China and Russia will likely keep increasing their deployed nuclear forces. Over the long term, the U.S. industrial base could be expanded—indeed, various expansion efforts are underway—but, realistically, the current nuclear modernization program cannot be accelerated by much, if at all.

Paying for that program, let alone fielding a larger nuclear force, will come with real trade-offs: either Washington will have to make significant cuts to nonnuclear programs or it will have to increase the overall defense budget substantially. The U.S government expects to spend $355 billion on nuclear modernization, out of a total anticipated expenditure of $660 billion on nuclear weapons, between 2023 and 2032.^[34] Not only are estimated program costs still growing, but those figures almost certainly underestimate the true cost.^[35] Based on historical cost-growth data, the U.S. Congressional Budget Office’s best estimate for actual expenditure on nuclear weapons between 2023 and 2032 is $756 billion, 15 percent higher than the official estimate.^[36] The development of a new nuclear-armed sea-launched cruise missile (SLCM-N), which the Biden administration committed to in 2024, will only increase the price tag.

For its part, the U.S. Department of Defense has acknowledged that cost growth to date, particularly in the Sentinel ICBM program, will require cuts elsewhere.^[37] By ignoring the likelihood of further cost growth, however, the Pentagon is underplaying the magnitude of the required cuts. As one U.S. Air Force official noted, because extra money will not need to be spent on Sentinel in the next five years, a decision on trade-offs is still “far down the road.”^[38] Punting on such a decision is the height of bureaucratic convenience, but it is hardly strategically optimal because it will lead the United States to waste money on programs that will be cut later.

Though the need for modernization is real, so are its costs and the limitations of the United States’ defense industrial base. Those costs and limitations should militate against further building up of the U.S. nuclear force.

The Costs of Counterforce: Escalation Risks

The dangers of counterforce targeting do not end with an arms race. Chinese or Russian fears that their nuclear forces are vulnerable to U.S. attacks (whether well-founded or otherwise) could spark otherwise avoidable escalation.^[39]

As proponents of counterforce emphasize, China and Russia value their nuclear forces greatly and worry about their survivability. Especially during a grave crisis or a conventional conflict, if an adversary believed that the United States could significantly degrade its nuclear force, it could resort to extreme measures—including using nuclear weapons—to protect that force. The most extreme possibility is that Russia could launch large-scale counterforce strikes against the United States before the United States could do the same to Russia. (China currently lacks the capabilities for such an operation.) More likely, China or Russia could use nuclear weapons in a limited way, to terrify the United States into ending a war rapidly or at least desisting from whatever behavior Beijing or Moscow found threatening.

Less dramatic actions could catalyze escalation, culminating in nuclear use.^[40] For example, China or Russia could issue nuclear threats as a prelude to nuclear use. Alternatively or additionally, either country could take steps to make its retaliatory capability more survivable. For example, Chinese or Russian leaders could pre-delegate launch authority down the chain of command so that, if they were killed in a U.S. nuclear strike, their subordinates could retaliate. They could also order an increase in the alert level of their nuclear force by, for example, dispersing mobile delivery systems. Each of those steps would risk further escalation. Pre-delegating launch authority, for example, would increase the chance of unauthorized nuclear use by Chinese or Russian officers entrusted with such authority. Nuclear threats or the dispersal of nuclear forces could prompt countervailing actions by the United States, including perhaps counterforce strikes if U.S. leaders believed that a nuclear war had become imminent.

The United States’ embrace of counterforce targeting exacerbates all those escalation pressures. In practice, one probable element of its doctrine could prove especially dangerous. In 2023, a study group convened by Lawrence Livermore National Laboratory publicly endorsed “limited counterforce strikes”; although left undefined, such strikes could presumably involve attacks against, for example, a handful of Russian ICBM silos.^[41] Because the study group included various former officials with extensive knowledge of U.S. nuclear strategy, it seems likely such attacks are part of U.S. nuclear plans. Yet it is difficult to imagine a better way to convince an adversary that its nuclear forces are in jeopardy and that it has no choice but to escalate. Indeed, even the study group had to acknowledge that “counterforce, either as first use or retaliation, can . . . muddle the message sent through limited use, potentially communicating that the attacking side is attempting a disarming attack.”^[42] In theory, the United States could plan for large-scale counterforce operations for the purpose of damage limitation, while abandoning the option of launching limited counterforce strikes for coercion. But the fact that the United States appears not to have done so suggests just how difficult it is to manage the escalation risks inherent in counterforce targeting.

The Overstated Case for Damage Limitation

The goal of damage limitation through counterforce targeting has intuitive appeal, but there is a problem: it would be impossible for the United States to limit the devastation it would suffer in a nuclear war to the extent that a U.S. president would view such a war as being significantly more desirable than one without damage-limitation strikes.^[43] For one thing, on the rare occasions that advocates for damage limitation offer detailed technical arguments, they make assumptions that, from the U.S. perspective, are unreasonably optimistic. That deficiency is most evident in discussions about the feasibility of tracking and destroying an adversary’s dispersed mobile ICBMs. The political scientists Kier Leiber and Daryl Press argue that, for adversaries, “the task of securing nuclear arsenals against attack [is becoming] much more challenging.”^[44] But they implicitly assume that North Korea (the state on which they focus) would operate its missile force in ways that are far from optimal.^[45] Specifically, they wave away the possibility that North Korea could shield dispersed mobile missiles from the United States’ overhead intelligence, surveillance, and reconnaissance (ISR) assets, such as satellites and drones. In fact, potential shelters are ubiquitous—bridges, tunnels, certain types of camouflage, or anything else that could block the line of sight between a North Korean missile and a U.S. ISR asset. The possibility that missiles could move between shelters in the intervals when ISR assets were not overhead would quickly erode whatever knowledge the United States had about the locations of North Korean ICBMs during a conflict.^[46] Moreover, as Leiber and Press themselves acknowledge, even if their conclusions about North Korea are correct, they are not directly applicable to China or Russia “because those countries are larger, with different topography, bigger arsenals, and more modern defenses.”^[47]

Second, even if the United States could destroy dispersed mobile missiles, an adversary might still be able to launch nuclear weapons, particularly silo-based ICBMs, after detecting an incoming nuclear strike but before the incoming warheads detonated.^[48] Russia has long had such a launch-under-attack capability and, according to the U.S. Department of Defense, China is well on the way to attaining one.^[49] But advocates of damage limitation often ignore the implications of those capabilities—and, on the rare occasions they do mention launch under attack, it is usually only to note that the issue lies outside their paper’s scope.^[50]

Third, China and Russia always have the option of deploying more nuclear weapons. Building nuclear weapons is expensive, as the United States is all too aware. At the end of the day, however, possessing more of them than an adversary can destroy is the most reliable way for a state to ensure the survivability of its retaliatory capability. In fact, given the value that China and Russia place on their nuclear forces, they will probably expand those forces if necessary to prevent vulnerability—just as China is already doing.

To be fair, some proponents of counterforce targeting recognize those limitations. The Lawrence Livermore Study Group, for example, acknowledges that counterforce strikes “would likely not enable the targets of nuclear retaliation to escape very large-scale death and destruction of a scale and character that U.S. and allied leaders would deem unacceptable.”^[51] In response, it points to the moral value of saving some lives “even amidst catastrophe.”^[52] That benefit, however, should be offset against the escalation risks generated by counterforce targeting. By making nuclear war—particularly large-scale nuclear war—more likely, the net effect of counterforce targeting is to risk lives.

U.S. Allies: Caught Between Plymouth Rock and a Hard Place

For the United States, the costs of extending deterrence are high—but so are the costs of not extending deterrence. On the one hand, if the United States abandons its allies (either by formally renouncing defense agreements or by engaging in behavior that leads allies to conclude those agreements are worthless), then it will lose the security, diplomatic, and economic benefits of alliances and spark a wave of proliferation by its partners. On the other hand, continuing to extend deterrence requires the United States to risk war, including nuclear war, on behalf of an ally and is likely to fuel arms racing.

From the United States’ entry into World War II to the beginning of Trump’s second term, successive American administrations reached the conclusion that the benefits of alliances outweighed the risks. Indeed, they sought not only to deter the adversaries of U.S. allies but also to assure those allies that their adversaries were deterred. The core challenge with assurance is political: convincing allies that if a conflict escalated, the United States really would trade Washington for Warsaw (or, indeed, a different American city for its alliterative foreign counterpart). Because persuading allies of the United Sates’ political will is difficult, assurance efforts have tended to focus on capabilities. Historically, one important element of assurance has been maintaining “nuclear forces that were perceived as ‘second to none,’” as former Undersecretary of Energy for Nuclear Security Linton Brooks has put it.^[53] In recent years, both U.S. and allied officials have often stressed, at least in private, that they view the United States’ maintenance of credible counterforce capabilities as important for extended deterrence—implying that a U.S. nuclear buildup is increasingly important for assuring allies. As a result, if Washington continues with assurance efforts, it is likely to feel pressure to increase the size of its nuclear force and to maintain counterforce targeting.

Of course, Washington will not necessarily continue those efforts. With Trump back in office, U.S. grand strategy, including the future of its alliances, is now uncertain in a way it has not been for over eighty years (relations between the United States and its allies were tense during the first Trump administration but significantly less fraught than today). During the 2024 election campaign, Trump recounted that he previously told one ally that he “would encourage” Russia “to do whatever the hell they want” to “delinquent” NATO members.^[54] In recent months, the president reminded NATO members of his previous threat that “if you’re not going to pay, we’re not going to defend,” before stating that allies “should be paying more.”^[55] Perhaps even more worrying for U.S. allies is that Trump called the very basis of NATO’s collective security arrangement into question by musing that “if the United States was in trouble and we called [U.S. allies], we said we got a problem . . . do you think they’re going to come and protect us? . . .  I’m not so sure.”^[56] He has made what two allied countries, Canada and Denmark, have reasonably interpreted as military threats against their territorial integrity (the latter in connection with Greenland).^[57] Meanwhile, European allies are concerned that Trump is sacrificing their interests—not least in Ukraine—in his attempt to forge stronger relations with Russia, which was exempted from the tariffs his administration applied to almost every other state in 2025, including all U.S. allies. The net effect is that nervousness among Asian and particularly European allies has escalated almost into panic, leading them to openly debate whether they should acquire their own nuclear arsenals.^[58] 

Against that backdrop, there are two reasons why, even if the United States tries capabilities-based assurance in the future, it is likely to fail. First, possessing a capability does not persuasively demonstrate the will to use it. Spending money on maintaining or acquiring some capability, especially in the face of significant domestic opposition, is not cost-free for the United States, making it more than just cheap talk. Even so, overcoming domestic barriers does not lock the United States into using that capability on behalf of an ally—as allies are all too aware, even if the exigencies of alliance solidarity preclude them from openly saying so. Indeed, U.S. allies are not so credulous that they believe putting more nuclear capabilities at Trump’s disposal will make him more inclined to defend an ally.

Second, even if increasing the overall size of the United States’ nuclear arsenal (or, indeed, acquiring SLCM-N or some other new capability) does assure allies, such effects will be only temporary. Any such benefits will derive from the fact that the United States is doing something more, rather than the specific nature of that something. As the psychologist Steve Kull wrote in 1988, “the value of nuclear weapons [as political symbols] lies as much in the act of building the weapons as in actually having them.”^[59] The flip side is that “once weapons are completed they lose their symbolic saliency.”^[60] Indeed, it is wishful thinking to believe that today’s entirely real assurance gap just happens to be the exact size and shape of whatever new American capability allies want. Even if they applaud the United States for fielding it, they will simply pocket the gain and move on to quietly insisting that U.S. credibility now hinges on whatever new nuclear weapon is flavor of the month within the Beltway. That nervousness among allies is fully understandable; extended deterrence requires an ally to place a tremendous amount of trust in the hands of the United States, which has betrayed that trust. However, sympathy for allies’ real predicament does not obviate the reality that their appetite for assurance will never be sated by American capabilities alone.

A Better Way Forward: Optimal Deterrence

The United States now faces a future fraught with growing risks of proliferation, arms racing, and escalation. Although Beijing and Moscow share the blame for this sorry situation (which poses as much danger to them as it does to Washington), the prospects for arms control are bleak. Even so, Washington is not a hapless bystander. It can change its policies—and it should do so where current policies are part of the problem. Unilateral action is no panacea, but it offers the best way forward.

Targeting Policy

The first-order question facing the United States is not whether to increase the size of its nuclear forces in response to the dictates of its current targeting policy; it is whether that policy best serves U.S. interests. In fact, a better alternative exists: the United States should target adversaries’ conventional military forces and war-supporting industry (CMI) assets.^[61] It should not deliberately target its adversaries’ nuclear forces, nuclear C3I systems, leadership, or population.^[62] The United States could effectively implement a CMI strategy—if necessary, against both China and Russia simultaneously—without a larger nuclear arsenal or additional low-yield capabilities.^[63]

The United States generally considers CMI assets to be legitimate military targets.^[64] Indeed, it almost certainly already targets them under current policy. As such, CMI targeting creates no additional legal complications compared to current policy. Given the infeasibility of meaningful damage limitation, adopting CMI targeting would not undermine deterrence, including extended deterrence, or the United States’ ability to respond to a potential deterrence failure.

The most important test case for any targeting policy is a regional conventional war—sparked, for example, by a Chinese invasion of Taiwan or a Russian invasion of the Baltic states—that the United States was losing. In such a scenario, the best (or perhaps least bad) theory for nuclear use would be coercion. In other words, the United States would start by using nuclear weapons in a limited way to terrify its adversary into backing down or compromising rather than risking all-out escalation. A necessary (but in no way sufficient) condition for success in that strategy is a targeting policy that gives the president limited nuclear options that could credibly be executed early in a nuclear war, as well as less limited options held in reserve during the early phases of conflict.

Conventional military assets are the most credible targets for limited nuclear use. Initially, the United States could attack assets—such as Chinese surface ships approaching Taiwan or isolated Russian military bases—whose destruction would signal the United States’ willingness to escalate further while causing, at most, limited harm to civilians. If necessary, Washington could escalate further by attacking facilities—ports, airfields or transportation nodes, for example—directly involved in supporting the adversary’s military operations. The effect of such attacks could be to degrade the adversary’s ability to fight a conventional war; their purpose, however, would be coercion. That distinction is crucial for force sizing because war plans premised on coercion—using violence to induce certain behavior by an adversary—require significantly fewer nuclear weapons than those that aim to achieve military objectives through brute force.^[65] War-supporting industry, such as electricity, oil, and metal production, could be attacked if there were further escalation. In the event of a full-scale nuclear war, attacks against CMI targets would essentially destroy an adversary as a functioning state; indeed, so would large-scale nuclear strikes under the United States’ current targeting policy, not least because many counterforce targets, such as leadership bunkers, are in or around cities.^[66] As such, the United States could inflict a degree of pain through CMI targeting that would far outweigh whatever gains the adversary sought through aggression.

Some defenders of the status quo disagree, arguing that because Beijing and Moscow value their nuclear forces so much, counterforce targeting is necessary for effective punishment.^[67] Indeed, Beijing and Moscow no doubt believe their nuclear forces are useful for deterring threats, compelling concessions (especially in Moscow’s case), and conferring status. But to debate precisely how much China and Russia value their nuclear forces is to miss the point. For targeting policy, the key question is whether U.S. attacks against an adversary’s nuclear forces and CMI targets would inflict meaningfully greater costs on China and Russia than attacks on CMI targets alone.

Clearly, the answer is no. The costs of a large-scale nuclear war to China and Russia would be unimaginably huge whether or not the United States attacked their nuclear forces. Hundreds of millions of their citizens would be dead or dying. Their economies, including food production and other basic services, would be devastated. Their conventional military forces would be annihilated. Following an all-out nuclear exchange involving CMI strikes by the United States and Chinese or Russian nuclear retaliation, any nuclear weapons that China or Russia had not used would have little value for self-defense because there would be little left to defend. Threats to use those weapons for offensive purposes would lack any credibility without being backed by the conventional forces needed to take and hold territory, and national prestige would not be a major concern for any Chinese or Russian leaders who happened to survive the apocalypse.

Escalation Management

Some of the benefits of CMI targeting—principally the absence of pressure on the United States to compete in an arms race—would materialize whether or not China or Russia believed the United States had really abandoned counterforce targeting. The realization of other benefits, such as reduced escalation risks, would require the United States’ renunciation to be credible.

At least initially, China and Russia would surely doubt U.S. leaders’ statements that the United States no longer targeted their nuclear forces. Over time, however, evidence to confront their skepticism would emerge. Specifically, the fact that the United States was not engaging in an arms race—that it was not building up its nuclear force in response to a Chinese and potential Russian buildup—would help persuade them that the country truly had abandoned counterforce targeting.

Especially in a crisis, a conventional conflict, or a limited nuclear war, the United States should refrain from taking actions that could persuade an adversary its nuclear force was in jeopardy. To that end, in addition to not conducting limited counterforce strikes (which would automatically be precluded by a policy of CMI targeting), it could

• not conduct airborne ISR operations in the patrol areas of Chinese or Russian ICBMs;
• not conduct anti-submarine warfare operations in the bastions for Chinese or Russian SSBNs or limit such operations to reduce the threat to those SSBNs (by, for example, not trailing them); and
• not conduct attacks (nuclear or nonnuclear) on C3I assets used for both nuclear and nonnuclear operations.^[68]

Each step would help manage escalation, but it would come with the downside of reducing the United States’ ability to prosecute a conventional war effectively. Credibly evaluating the trade-offs requires classified information; accordingly, U.S. planners and operators should consider each step and its trade-offs in developing a strategy for fighting a conventional war in a way that minimizes the risks of nuclear escalation.

Assuring Allies

For now, the tactical question of how to assure allies is moot, having become subservient to the strategic question of whether alliances serve the U.S. national interest to begin with. If the United States and its allies part ways, assurance will become entirely irrelevant.

If, however, the United States recommits itself to its alliances during this administration or afterward—as it should—it will have to somehow rebuild trust with its allies. That immensely difficult task would be made yet more challenging by budget constraints and the need to prepare to fight wars in both Asia and Europe. With that in mind, two principles should guide future efforts to assure allies while mitigating the risks of proliferation and arms races.

First, the United States should maintain and acquire whatever capabilities are necessary for extended deterrence, but it should apply a careful cost-benefit test to any additional steps—that is, steps that are unnecessary for deterrence but potentially useful for assurance. In applying that test, Washington should recognize the limited and temporary benefits that new hardware offers in terms of assurance; upgrading alliance “software” through better coordination, consultations, and communications is cheaper and more durable. If building more nuclear weapons to meet counterforce-targeting plans is an example of an assurance measure that fails the cost-benefit test, the Washington Declaration—a U.S.-South Korea joint statement from 2023—passes it. That declaration did not eschew capabilities-based assurance—the United States promised to “further enhance the regular visibility of strategic assets to the Korean Peninsula.” However, such operations were only one part of a broader strategy that included establishing a new Nuclear Consultative Group to “discuss nuclear and strategic planning.”^[69]

Calling on the United States to ensure it has the capabilities needed for extended deterrence is not just empty rhetoric. Today, the United States lacks those capabilities—but the main gaps are in the conventional and not the nuclear domain. NATO’s conventional weakness in and around the Baltic states and China’s growing strength in the West Pacific increase the likelihood of a conventional war that could turn nuclear. Ultimately, addressing those weaknesses is a better way to reduce nuclear risks than building up the United States’ nuclear arsenal.

Second, even in the best case, allied proliferation will be a serious risk going forward. (In the worst case, it will be a reality.) Adding to that risk, some allies probably believe that if they acquired nuclear weapons, the United States would continue to provide security guarantees. They perhaps also hope that the U.S. government would find some way to exempt them from the sanctions it would be required to impose under domestic legislation were they to develop nuclear weapons. (Some laws provide presidential waiver authority; in other cases, sanctions could be stopped only through new legislation.^[70])

To ensure allies believe that proliferating will incur real costs, the United States should present them with a clear choice: they can have their own nuclear weapons or rely on the United States’, but not both. The United States cannot ultimately prevent an ally that is determined to proliferate from doing so. But in response, it can and should withdraw security guarantees. And Washington should be clear that it will not waive sanctions and will oppose or veto any attempt by Congress to nullify them.

Allies’ serious doubts about the United States’ willingness to defend them would significantly lessen the leverage that Washington can wield through a threat to withdraw security guarantees; after all, because of those doubts, they may feel they have little to lose by going nuclear. That said, proliferation would still come with serious downsides for allies. Chief among them is the risk of falling victim to a preventive attack. In response to evidence that Poland, say, was seeking the bomb, Russia could launch large-scale air strikes against Polish nuclear facilities. Moreover, proliferation would be economically expensive, given the direct costs and sanctions it could trigger; domestically controversial, especially in democracies; and reputationally risky, particularly for states that have piously preached nonproliferation for decades. Accordingly, allies could still judge a less-than-watertight U.S. security commitment to be less risky than proliferation.

Withdrawing a security guarantee from a European ally would be particularly fraught for the United States because it would effectively turn NATO into a two-tier alliance; most member states would enjoy U.S. protection, but new proliferators would not. But if the United States is to use security guarantees to prevent proliferation—as it has sought to do for decades—it should ultimately be prepared to accept that outcome. Moreover, extending security guarantees to an ally with nuclear weapons is particularly dangerous. Emboldened by its nuclear arsenal, the ally could be more willing to embark on some course of action that could start a conventional or even nuclear war with China, North Korea, or Russia, and then expect the United States to step in and finish the job.^[71]

Recommendations for U.S. Policy

The United States’ current nuclear strategy—centered on counterforce targeting—is likely to spark a tension-generating arms race and exacerbate escalation risks during a conflict. It is possible, however, to reduce those dangers without compromising the security of the United States or its allies. To that end, Washington should implement an alternative strategy of optimal deterrence by

• ceasing, and declaring it has ceased, the targeting of adversaries’ nuclear forces, nuclear C3I systems, and leadership, and instead focusing exclusively CMI assets, which it already targets;
• weighing the pros and cons, in war planning, of not conducting conventional or nuclear operations that could convince an adversary that the United States was planning large-scale damage-limitation attacks on its nuclear forces; and
• continuing ongoing modernization programs, including the development of the Columbia-class SSBN and B-21 bomber, that are needed to preserve the survivability and reliability of U.S. nuclear forces.

The United States should also rebuild its relationships with allies, both because alliances bolster U.S. security and because abandoning allies would risk sparking a wave of proliferation. That said, the United States should not go back to allowing allies to practically dictate U.S. force requirements in the interests of assurance. Instead, to ensure the sustainability of any future alliances, the United States should

• acquire the capabilities—including nonnuclear capabilities—required to deter allies’ adversaries;
• apply a clear-eyed cost-benefit test to potential assurance measures that are unnecessary for deterrence; and
• make clear to allies that if they acquire nuclear weapons, they will lose U.S. extended deterrence guarantees and will be subject to the sanctions mandated by U.S. domestic law.

Conclusions

For the last fifty years, the full implications of counterforce targeting, which creates a tight link between U.S. requirements and adversaries’ capabilities, were obscured by the United States’ and Russia’s willingness to settle for parity in long-range nuclear forces. Parity did not end strategic competition—the United States continued to enhance its ability to conduct counterforce strikes through improvements in missile accuracy and ISR capabilities, for example—but it did halt the quantitative arms race. The growth in China’s nuclear arsenal has created a genuine inflection point for the United States. If the United States wishes to continue with counterforce targeting, it will have to build up indefinitely.

A new arms race would be expensive and increase tensions, thus increasing the likelihood of a future war. Officials and analysts, however, tend to speak fatalistically about the prospect of competition in offensive nuclear forces. They blame other states for starting an arms race and, in so doing, imply that Washington cannot but compete; as Heritage Foundation Fellow Robert Peters puts it, “the United States has failed to keep pace . . . . While the autocrats in Beijing, Moscow, Pyongyang, and potentially Tehran are running in a nuclear arms race, the United States is jogging in place.”^[72] U.S. officials tend speak about a potential buildup as if the United States has no agency; as then Principal Deputy Assistant Secretary of Defense Vipin Narang put in 2024, “adjustments to the number of deployed capabilities may be necessary.”^[73]

But the United States does have agency. Adopting counterforce targeting—and everything that entails—was a choice. Targeting conventional military forces and war-supporting industry offers a superior alternative.

Changing policy will encounter considerable opposition, including from the Pentagon, Capitol Hill, and allies. Critics will make the standard arguments in favor of counterforce targeting—that it is needed for damage limitation, effective punishment, and assuring allies. Yet those arguments do not stand up to scrutiny. Meaningful damage limitation is impossible. CMI targeting could impose far greater costs than an adversary could gain through successful aggression—and a counterforce-targeting policy is simply not an effective way to address allies’ entirely reasonable concerns about the reliability of the United States as a security guarantor.

The risks brought by counterforce targeting—escalation and arms races—are all too real. “We know how to win these races,” the U.S. special presidential envoy for arms control declared in the first Trump administration, “and we know how to spend the adversary into oblivion.”^[74] But in reality, the prospects for winning a three-way arms race by spending Russia and China into oblivion are poor. The prospects for winning a nuclear war are even worse. The United States should face these facts sooner rather than later.

Acknowledgments

I thank Anna Bartoux for research assistance. For their extremely helpful comments on earlier drafts, I gratefully acknowledge Jerry Brown, Corey Hinderstein, Ankit Panda, George Perkovich, Paul Stares, Pranay Vaddi, and the participants at a workshop, which I thank Natalie Caloca for helping to organize. Naturally, I am exclusively responsible for this report’s contents.

About the Author

James M. Acton holds the Jessica T. Mathews chair and is codirector of the Nuclear Policy program at the Carnegie Endowment for International Peace. Acton is currently writing a book on the nuclear escalation risks of advanced nonnuclear weapons and how to mitigate them. His work on that subject includes the International Security article “Escalation Through Entanglement.” Acton has published in the New York Times, the Wall Street Journal, the Washington Post, Dædalus, Foreign Affairs, Foreign Policy, and Survival. He holds a PhD in theoretical physics from the University of Cambridge.