The void that suits everyone
Posted on: 16 February 2026
On 5 February 2026, the New START treaty expired. The last legally binding agreement limiting the nuclear arsenals of the United States and Russia is now gone. For the first time in over half a century, there is no treaty capping the strategic warheads of the two powers that hold roughly 85% of the world's nuclear weapons. The news drifted through the media cycle for a few days, wrapped in the customary alarmed tones, then vanished into background noise. Yet what happened is neither a diplomatic accident nor a negotiating oversight. It is the product of a convergence of incentives in which each of the three principal nuclear actors, the United States, Russia and China, finds in the institutional void precisely what it needs.
The treaty, signed in 2010 by Obama and Medvedev, limited each side to 1,550 deployed strategic warheads on a maximum of 700 delivery vehicles: intercontinental ballistic missiles, submarine-launched missiles and heavy bombers. It included a system of reciprocal inspections and data exchanges that experts considered the operational heart of the agreement: not so much the numbers, but the predictability. Knowing what the other side has, where it keeps it, how it moves it. This infrastructure of transparency crumbled well before the formal expiration date. On-site inspections stopped in 2020 due to Covid and never resumed. In February 2023, following the invasion of Ukraine, Putin suspended Russia's participation in the treaty, formally maintaining the numerical limits but eliminating any verification mechanism. The United States responded by ceasing its own data sharing. For three years, New START was a shell: the numbers on paper remained, but nobody could verify they were being honoured.
The dynamic that produced this outcome is more instructive than the expiration itself. Each actor had precise structural reasons not to prevent the void, while publicly declaring a desire to fill it.
For Washington, New START had become an increasingly uncomfortable straitjacket. The US nuclear triad modernisation programme, which the Congressional Budget Office estimates will cost $946 billion over the next decade with a 2026 annual budget of approximately $60 billion, is designed for a world in which America must maintain simultaneous deterrence against two peer nuclear adversaries. China, which the Pentagon estimates currently possesses over 600 operational warheads on a trajectory towards 1,000 by the end of the decade, has never been party to any nuclear limitation treaty. Trump put it with characteristic bluntness in a January interview with the New York Times: "If it expires, it expires. We'll just do a better agreement." He added that he would want to "get a couple of other players involved also." The American position is clear in its calculated ambiguity: no rush to bind itself again, because in the meantime the hands are free to adjust the arsenal to the new two-front reality.
For Moscow, the expiration offered an opportunity to appear responsible without actually being so. In September 2025, Putin proposed continuing to observe New START's limits for one year after expiration, provided the Americans did the same. The gesture was received positively by Trump, who called the idea "pretty good." But look at the mechanism: Russia proposes a voluntary commitment, without inspections, without data exchange, without any instrument that would allow verification of compliance. Lavrov confirmed this to the Duma on 11 February 2026, stating that Putin's "moratorium" would remain in force "as long as the US doesn't exceed these limits," adding that Russia would act "in a responsible and balanced way on the basis of analysis of US military policies." Translated from the language of diplomacy: we will do what suits us and say it was what we promised. The absence of verification renders any commitment unverifiable by definition. It is the form of restraint without the substance of constraint.
For Beijing, every day without a treaty is a day gained. China has never been part of New START and categorically refuses to participate in any trilateral negotiation. The reason is purely arithmetic: with approximately 600 warheads against America's 3,700 and Russia's 4,300 ready for use, according to the most recent estimates, China has no interest whatsoever in freezing its arsenal at such a marked level of inferiority. Washington's insistence on a treaty that includes Beijing, however understandable from a strategic viewpoint, functions in practice as a perfect stalling mechanism: the US conditions a new agreement on Chinese participation, China refuses, and in the meantime no agreement binds anyone. Time works for Beijing, which is building new silo fields for intercontinental ballistic missiles, has deployed a complete nuclear triad with land, sea and air capabilities for the first time, and is developing hypersonic technologies that existing treaties never even contemplated.
The informal agreement negotiated in Abu Dhabi on the sidelines of the Ukraine talks on 5 February, the very day of expiration, reveals the true nature of the situation better than any official statement. According to Axios, Trump's envoys Steve Witkoff and Jared Kushner discussed with Russian officials a "handshake" to observe New START limits for at least six months while negotiating a successor. A US official specified that the agreement could not be legally formalised. Kremlin spokesman Peskov commented that "it's hard to imagine any informal extension in this sphere." The negotiations took place without the active participation of arms control experts from the State Department: Under Secretary Thomas DiNanno was in Geneva for a UN disarmament conference. The most delicate architecture in the international system, the one governing the very possibility of human survival, is being handled on the margins of other negotiations, by people whose primary mandate lies elsewhere.
The most relevant historical precedent is the SALT II treaty, signed by Carter and Brezhnev in 1979 but never ratified by the US Senate following the Soviet invasion of Afghanistan. For seven years, from 1979 to 1986, the United States and Soviet Union informally respected an agreement that had no legal force. Carter declared he would observe the terms as long as Moscow did the same; Brezhnev issued an analogous statement. Reagan, who had called SALT II "fatally flawed" during his campaign, promised in 1982 not to do anything to undercut it. The system held, more or less, until 1986 when Reagan abandoned it, citing Soviet violations. But that precedent operated in a radically different context: a bipolar competition with only two relevant actors, relatively symmetrical arsenals, and a long track record of structured negotiation. Today's situation is triangular, asymmetric and devoid of any consolidated negotiating infrastructure. But there is an even deeper difference, and it concerns time. In 1979, an intercontinental ballistic missile took roughly thirty minutes to reach its target: thirty minutes to verify the alarm, confirm the launch, consult the chain of command, decide. The entire framework of classical deterrence presupposed that margin. Hypersonic weapons have compressed it to under ten minutes. Cyber warfare systems can compromise early warning networks in seconds. The integration of artificial intelligence into detection systems introduces a level of complexity where human decision-making risks becoming not a mandatory step but a bottleneck to be eliminated for reasons of speed. The informal SALT II worked, with all its limitations, in a world where reaction time still allowed reflection. Using it as a reassuring precedent today means ignoring not only that the players are three instead of two, but that the game itself moves at a speed incompatible with the handshake as a security mechanism.
According to a January 2026 poll, 91% of Americans believe the US should negotiate a new agreement with Russia to maintain or further reduce nuclear limits. 85% of those who voted for Trump think the president should accept the Russian proposal. There is public consensus, there is declared interest from both sides, yet the treaty expired without anyone doing anything concrete to prevent it. When all the apparent conditions for an agreement are present and the agreement does not materialise, the problem is not in the conditions: it is in the real incentives.
Those who operate in the real world, who make decisions with measurable consequences, should examine this episode with surgical attention. The trajectory is clear: the world is moving towards a three-way nuclear competition without institutional guardrails, where deterrence relies entirely on each actor's ability to correctly assess the intentions of the other two. It is a system that works perfectly as long as nobody gets a calculation wrong. The history of the twentieth century suggests that miscalculations in nuclear crises are not rare exceptions but near statistical certainties over a sufficiently long time horizon: the Cuban Missile Crisis in 1962, the Petrov incident in 1983, the Able Archer exercise the same year. In each of these cases, catastrophe was averted not by treaties but by luck or by the individual judgement of single persons who chose not to follow protocol. That is not a foundation on which to build the security of an entire century.
The point is not that anyone is planning a nuclear conflict. The point is that the institutional void suits everyone in the short term and suits no one in the long term. It is the classic structure of a collective action problem where every rational actor, pursuing its immediate interest, produces an outcome that harms everyone, including itself. But this incentive analysis contains its own blind spot that must be declared: it assumes that actors always calculate, that decisions are made by clear minds under conditions of reasonable information. Nuclear history demonstrates the opposite. In moments of acute crisis, rationality does not calculate: it contracts, freezes, defaults to protocol or to panic. Petrov in 1983 did not conduct "incentive analysis" when the Soviet early warning system signalled five incoming American missiles: he had an intuition that contradicted everything protocol demanded he do, and he chose not to launch the retaliatory strike. Had he followed the rules, the world as we know it would have ended that night. When institutional protocols erode, when treaties expire and inspections cease, it is not only predictability that vanishes: it is the infrastructure that allows rationality to operate under extreme pressure. What expands is the space in which decisions are driven by paranoia, by technical error, by the fear of appearing weak. And in that space, the calculus of incentives is no longer of any use.
Those who only look at the news see a technical expiration. Those who look at the structure of incentives see a system that designed itself to produce precisely this result. Those who look deeper still see that the system does not need irrational actors to fail: all it needs is a single moment of crisis in which the time available to decide is less than the time required to think. Today that margin has narrowed as never before, and the architecture that compensated for it is gone.