A new nuclear arms race is beginning. It will be far more dangerous than the last one

Like Toto in The Wizard of Oz, at their 1985 summit in Geneva President Ronald Reagan and the Soviet leader Mikhail Gorbachev pulled back the curtain to reveal the truth behind the terrifying spectre of nuclear war, which their countries were spending hundreds of billions of dollars to prepare for. "A nuclear war cannot be won," they jointly stated, and "must never be fought." They omitted the inescapable corollary of those first six words: a nuclear arms race also cannot be won.

Still, the statement, almost unique among government declarations for its blunt truthfulness, strengthened the case for the arms control and nonproliferation undertakings that followed. Decades of agonisingly difficult negotiations built up a dense structure of treaties, agreements and even a few unilateral moves dealing with offensive and defensive nuclear weapons of short, medium and long range, with provisions for testing, inspections and an overflight regime for mutual observation. Often the two sides would only give up systems they no longer wanted. Frequently the language of the agreements was the basis of future friction. On the US side, the political price of securing Senate ratification of treaties could be extremely high.

But for all its shortcomings, arms control brought down the total number of nuclear weapons held by the two countries from 60,000 to roughly 11,000 today. (The exact number is classified.) Under the most recent treaty, New Start (strategic arms reduction treaty), signed in 2010 , each side is limited to 1,550 deployed weapons, with the rest in storage. By any accounting, that 80% drop (95% counting just deployed weapons) is – or was – a notable achievement.

Unfortunately, the past tense is correct, because since the US withdrew from the anti-ballistic missile (ABM) treaty in 2002 – thereby legitimising the unilateral renunciation of an agreement by one party if it no longer finds the restrictions to its taste – the other agreements have fallen one by one. In February 2026 – about 500 days from now – New Start, the last remaining brick in the edifice so painstakingly built, will expire, leaving the US and Russia with no restrictions on their nuclear arsenals for the first time in half a century.

With tensions among the great powers at a post-cold war high, a new nuclear arms race is beginning. This one will be far more dangerous than the first. It will be a three-sided race – now including China – and thus much more unstable than a two-sided one. And it will be amplified by the advent of cyberweapons , AI, the possible weaponisation of space, the ability to locate submarines deep in the ocean and other technological advances.

To appreciate the danger this represents, it is necessary to look back at the peculiar dynamics of a nuclear arms race and see the craziness that drives intelligent people in its grip to grotesque extremes. From 1950 to 1965 the US arsenal grew from its first few warheads to more than 30,000 – about five times as many as the Soviet Union had at the time. Its bible then and now has been the single integrated operational plan (Siop) – the multiservice plan for nuclear war. The Siop specifies the targets to be attacked and is based on a required level of confidence with which each one must be destroyed.

As recounted by Fred Kaplan in his brilliant history The Bomb , President Dwight D Eisenhower made one of Washington's early attempts to exert some control over the nuclear planning being done in its name.

In November 1960, he sent his science adviser, George Kistiakowsky, accompanied by another weapons expert, George Rathjens, to strategic air command (SAC) headquarters in Omaha, Nebraska, for a briefing. Rathjens came prepared with the name of a Soviet city similar to Hiroshima in size and industrial capacity and asked what weapons the Siop assigned to it. The answer was one 4.5-megaton bomb followed by three 1.1-megaton bombs –a lunatic total of more than 500 times the 15-kiloton bomb dropped on Hiroshima.

As outsiders to SAC looked more and more closely at its work, they were variously baffled, stunned and appalled. Asked how many Russians, Chinese and eastern Europeans would be killed in the all-out attack envisaged by the first Siop, SAC's answer was 275 million – counting only deaths from bomb blast but not those from heat, fire, smoke and radioactive fallout, because these could not be precisely calculated. Actual fatalities, therefore, would be many times greater. The population of the region at the time was 1.03 billion. Unthinkable as this was, fear of the Soviet Union, SAC's clout inside the military establishment and political momentum in Washington were so great that, administration after administration, the president and the Pentagon wrote guidance that SAC turned into monstrous plans.

Decades later, after several rounds of cuts, the number of weapons had come way down, but the level of overkill in the Siop was still bizarre. A review ordered by President Barack Obama revealed, for example, that the plan included several targets that were empty fields. According to US intelligence, these were designated as backup bases where Russian bombers could land if their primary bases had been destroyed. As Kaplan tells it, the official guidance required that "secondary bomber bases" be destroyed, so the Siop assigned not one but several weapons to each of these fields.

The otherworldliness of the anticipated war lies not only in its planning – it has operational echoes as well. USAF Gen Charles Boyd, a fighter pilot (and my late husband), served for a time in a posting in which his task would have been to deliver a nuclear weapon in the event of a war in Europe. He and his fellow pilots were each issued an eye patch lined with lead and instructed to put it on just before releasing their bomb. At the altitude at which their planes flew, their unprotected eye would be blinded by the flash of the explosion. They could then remove the patch and use that eye to fly. Losing sight in one eye was not a big concern, however, because the pilots had no doubt that this would be a one-way mission: there would be nowhere to land in western Europe in the throes of nuclear war.

Each year since 1947, the Bulletin of the Atomic Scientists has published a Doomsday Clock, set by a board of experts, as an easily understood assessment of the risk of global – mostly nuclear – catastrophe. In 1991, after the signing of the Start I treaty – the first to make deep cuts in the Soviet and US nuclear arsenals – the clock was set to 17 minutes before midnight. As of January 2024, it stands at 90 seconds , the closest it has ever been to the metaphorical moment of apocalypse. The board cited the widespread and growing reliance on nuclear weapons, the huge sums being spent to expand or modernise nuclear arsenals, and Russian threats to use nuclear weapons in the war in Ukraine.

A principal cause of concern is that after decades of relying on a very small nuclear deterrent, China is rapidly expanding its arsenal. It is now estimated to have 500 nuclear weapons , with plans to reach 1,000 by the end of the decade and perhaps numerical parity with the US and Russia by deploying 1,500 strategic weapons (those powerful enough to destroy cities and other distant targets) by 2035.

Washington can only guess at the motivation behind this decision. Because Beijing has never been willing to participate in arms control negotiations, the US government has little first-hand knowledge of its thinking about its nuclear forces and strategy. It could be preparing for war over Taiwan or seeking more broadly to establish hegemony over the Indo-Pacific. It could be responding to what it sees as American aggression. It could simply be taking the steps it feels are its due as a newly arrived great power now that it can afford to do so. Most likely there is a mix of motivations among different parts of the government.

Russia has modernised its traditional nuclear forces. In addition, in an angry speech in 2018, Putin unveiled several new nuclear weapons systems . He claimed that these were a response to the US withdrawal from the ABM treaty and its subsequent work on missile defences, which Moscow vehemently opposed in the belief that they would neuter its incoming missiles in a war. Ironically, despite enormous effort and expense over decades, US missile defenses have never been able to do that. The best they can do, even under test conditions arranged to make the task easier, is perhaps to intercept the equivalent of one or two North Korean missiles, but nowhere near a large Russian attack. So the US decision to leave the ABM treaty backfired in the worst way, adding little if any security while frightening and infuriating Russia. The new weapons announced by Putin include an intercontinental hypersonic glider, the trajectory of which can be altered during flight, a very fast nuclear-powered cruise missile of almost unlimited range, and an underwater nuclear torpedo that could span the Pacific. "No one has listened to us," Putin said. "You listen to us now."

Nonetheless, in President Joe Biden's first week in office, Russia and the US announced a five-year extension of the New Start treaty days before its expiration. Two years later, however, in a fit of anger over western support for Ukraine, Moscow announced that it was "suspending" the treaty. Both sides continue to observe the treaty's limits on weapons, but its critical provisions for verification via data exchange, notifications and on-site inspections are gone.

Later that year, Moscow took the further step of revoking its ratification of the 1996 comprehensive test ban treaty (CTBT ). Russia had ratified it in 2000, but in 1999, in a shocking step, the US Senate had rejected the treaty, although it was an American initiative that had been a national priority for years. As with New Start, however, Russia's deratification was a response to US support for Ukraine, not to its failure to ratify the treaty.

The CTBT has been ratified by 178 countries, though it cannot officially enter into force until the US, China and a few others join. In another angle on this strange story, President George HW Bush announced a nine-month moratorium on testing in 1992: new technology had provided the means to assure the safety and reliability of nuclear weapons without explosive testing. That moratorium has now stretched to 32 years, but the Senate has never been willing to reconsider the treaty. Nonetheless, the new norm has held. Except for those by North Korea , there have been no tests anywhere in the world since 1998.

Thanks to decades of exacting negotiations, the US has a good understanding of the doctrines, personalities and technical details involved in Russian nuclear planning. But two recent developments darken the picture. In 2022, days before the invasion of Ukraine, Putin and Chinese president Xi Jinping announced a "no-limits" partnership covering economic, geopolitical and security relations. China has in fact placed limits on what it will do, for example, in providing weapons for use in Ukraine, but the two countries' announced joint goal of ending American primacy in international affairs deeply concerns US leaders. Putin has also engaged in an unprecedented degree of nuclear sabre-rattling tied to the Ukraine war. As the invasion began, he put Russia's strategic weapons on heightened alert. He has since threatened to use tactical weapons (shorter-range weapons intended for battlefield use) if he thinks the west's support of Kyiv goes too far, and has moved some of these weapons into Belarus and ordered joint combat drills involving them. Most recently, officials have said that Russia's formal doctrine would be amended to lower the threshold for nuclear use. Notwithstanding all these steps, there is no doubt that Russia continues to be deterred by the unquestioned threat of an overwhelming US response to any nuclear use.

For its part, the US has embarked on a sweeping modernisation of its entire strategic triad covering new warheads, delivery vehicles (bombers, submarines, missiles), and support and command-and-control systems. Assuming no further overruns – unlikely – the cost to develop, build and operate the new systems will be at least $1.5tn. (For a sense of scale, 1tn seconds last about 32,000 years.) Several alternatives to replacing the land-based missile leg of the triad have been proposed for both technical and cost reasons as well as strategic ones. These Minuteman ICBMs (intercontinental ballistic missiles) are called "first-strike" weapons because the location of their fixed silos is well known and therefore in a war they must be fired quickly before they are attacked. For this reason, they are kept on high alert, making them particularly susceptible to accident or miscalculation. Thus they are vulnerable and destabilising. The anticipated cost of this one piece of the programme has ballooned by 81% since 2020 as the schedule has slipped. The Pentagon insists that it should proceed as planned nonetheless.

The combination of this enormous spending on upgrades and additions to the world's three largest nuclear arsenals; the high level of tension and mistrust among Russia, China and the US; and the anticipation of highly destabilising technological advances accounts for the Doomsday Clock's approach to midnight. Now there are signs that the second hand may have to be moved yet closer.

The US modernisation plan did not contemplate additions to the nuclear arsenal. It was to be an exchange of new and improved weapons for old – in some cases very old – ones, while keeping to New Start's limits on weapons and delivery systems. But New Start is almost 15 months from its expiration, with no replacement likely. Moreover, the US force was designed solely for use against Russia. A force powerful enough for that purpose was always deemed capable of dealing with a potential threat from China or any other state. Now China's apparent determination to quickly acquire a large strategic nuclear force and its new partnership with Russia dramatically alter the picture. The US faces not one but two nuclear "near peers".

The critical question is how to respond. Reflexive hawks have argued that the situation demands adding to the US arsenal. Until very recently the Biden administration, and many outside experts, argued that such thinking fundamentally misunderstands nuclear deterrence. "Nuclear deterrence isn't just a numbers game," said secretary of defence Lloyd Austin in December 2022. "In fact, that sort of thinking can spur a dangerous arms race." Six months later, national security adviser Jake Sullivan was more explicit: "I want to be clear here – the United States does not need to increase our nuclear forces to outnumber the combined total of our competitors in order to successfully deter them."

Rather than numbers, the meaningful measure of deterrence is whether, after absorbing even a joint attack, the US would still be able to inflict catastrophic damage on its enemies. That in turn depends heavily on what it chooses to target. Current US strategy, known as counterforce, aims at the opponent's nuclear forces, leadership, and military command and control structure. Other possibilities, such as targeting the infrastructure that holds a state together – its industry, ports, transport, finance, communication networks, government and conventional forces – might achieve the same result with far fewer warheads and a comparable number of civilian deaths.

By the summer of 2024, administration officials were signalling that rising nuclear risks could force the US to replace one-for-one modernisation with additions to its forces. This could lead to "nothing short of a new nuclear age", said Vipin Narang, a senior Pentagon official, on 1 August. "Absent a change" in Russian behaviour and China's nuclear trajectory, he added, we might have to look back on the period since the end of the cold war as "nuclear intermission". There are conflicting reports about whether the top-secret nuclear guidance signed by Biden last March reoriented nuclear planning around China. "We have repeatedly voiced concerns" about China's growing arsenal, said a national security council spokesperson in an intentionally non-informative comment, but "there is far more continuity than change" in the new guidance.

By raising the implicit threat level, the administration may be trying to push Moscow and Beijing toward different behaviour, or an internal bureaucratic battle may be ongoing or have been resolved in favour of a new policy. With former President Donald Trump's re-election, there is little doubt that the administration will respond aggressively to what it sees as an elevated threat from China. Trump's fourth and last national security adviser in his first term, Robert C O'Brien, argued this summer that "the United States has to maintain ... numerical superiority to the combined Chinese and Russian nuclear stockpiles ."

This would be almost a mathematical impossibility. Notwithstanding the supposed friendship between presidents Putin and Trump, after decades of US-Soviet and then Russian arms control agreements negotiated on the basis of parity, Russia would match any additions the US makes. China's force would be additional to that. Hence even equality in numbers – much less superiority – would be unattainable. Such a goal is a foolproof recipe for an arms race without end.

O'Brien also called for a return to nuclear testing – a truly baffling proposal. After conducting more than 1,000 tests, the US has little to learn from carrying out more. China, on the other hand, has conducted fewer than 50 and would leap at the opportunity to resume if the US were to bear the political opprobrium of breaking the global moratorium. It would not be long before other nuclear and currently non-nuclear states followed suit. The result would be a pointless loss in US national security and a sharp spur to global proliferation.

So it seems that the curtain has risen on the opening of a new arms race, one deeply destabilised by a third participant and a raft of new technology. There is little doubt as to how it will unfold and – God willing – end. Vast sums of money will be spent by each country to respond to worst-case assumptions about the others. The diversion of funds from domestic needs and the growing national debt burden will weaken all of them. This will continue until fear, among leaders and perhaps the public, turns wiser heads to diplomacy and the inching, difficult steps of negotiated arms control. Then additional vast sums will be spent dismantling what has been built. Darker possibilities of course abound, including accident, miscalculation or a "limited" nuclear war intentionally started or stumbled into in the delusion that it would not escalate to a global holocaust.

Those who bear the responsibility to protect the US must naturally consider worst-case assumptions about adversaries. But they must also bear in mind truths we now know about ourselves. When the US went on its nuclear building spree in the 1960s, it had no intention of destroying the planet even as it acquired more than enough nuclear weapons to do so. Even now, with so much more publicly known about this period, it would be hard to point to a single clear rationale for why the US created such an arsenal.

While it faces real threats, the US is still unquestionably the strongest of the three participants in this costly, dangerous and ultimately futile contest, with the opportunity to consider what might be done to interrupt it. A negotiated end to the war in Ukraine would help remove barriers to working with Russia. Even before that, there are possibilities. Gen Christopher Cavoli, the Nato supreme allied commander Europe and a fluent Russian speaker and student of that country, urges that Washington should make a concerted effort to revive lines of communication that enabled the US and the Soviet Union to survive the cold war: "We knew how to communicate verbally and nonverbally about our intentions in a way that gave predictability to the other side ... This was one of the principal things we used to ... achieve deterrence without significant risk ... We knew how to send signals to each other ... almost all of that is gone now."

Rose Gottemoeller, the chief US negotiator of New Start and former deputy secretary general of Nato, believes that even though Russia violated the 1987 intermediate-range nuclear forces (INF) treaty by deploying missiles with an illegally long range, an offer made by Putin in 2020 suggests the possibility, however small, of negotiating a new agreement on such missiles. China might be interested as well. Putin and Xi have reportedly discussed parallel moratoriums on INF missiles in Europe and Asia.

Attention needs to be paid to public opinion as well. The major steps to wind down the first arms race began under heavy public pressure that influenced leaders and legislators. Today, the climate crisis has replaced nuclear war as the main existential threat in the public mind, and funders of nongovernmental research and analysis have redirected their resources heavily in that direction even while the reasons for fearing nuclear war are, if anything, greater than they once were.

The Senate could also take a fresh look at the test ban treaty. Washington insiders will laugh at this idea, since the Senate has never shown the slightest willingness to reconsider its vote. But after 32 years of no testing, with proven means of assuring the reliability of US weapons without testing and with international monitoring stations in place capable of detecting the smallest test anywhere on the planet, there are no legitimate arguments against ratification. Moreover, 91 senators did not take part in the original vote. Though it would have no effect on US security, ratification would reverberate loudly worldwide. It would probably be followed by ratification by the other treaty holdouts and thereby boost global efforts to contain proliferation and rising nuclear risks.

This list is only suggestive. Others will have different ideas. The point is that even with formal arms control negotiations currently precluded by geopolitical tensions with both Russia and China, there are steps the US could take, without risk to its own security, to interrupt or possibly reverse the downward spiral that has begun into the depths of another full-fledged nuclear arms race.

An earlier version of this appeared in the New York Review of Books