No one can say with certainty, but the risk of nuclear weapons being used again is higher now than at any point since the Cold War. The Bulletin of the Atomic Scientists set its Doomsday Clock to 89 seconds to midnight in 2025, the closest it has ever been to catastrophe. That doesn’t mean a nuclear strike is imminent or inevitable, but it reflects a convergence of factors: expanding arsenals, collapsing arms control agreements, active wars between nuclear-armed states and their proxies, and new technologies that compress the time leaders have to make decisions.
What has prevented nuclear use since Hiroshima and Nagasaki is a combination of deterrence and something scholars call the “nuclear taboo,” a powerful global norm against using these weapons that has held for 80 years. Whether that norm can survive the pressures now building against it is the real question.
Why Nuclear Weapons Haven’t Been Used Since 1945
The standard explanation is deterrence: no country uses nuclear weapons because the retaliation would be devastating. That’s true, but it’s not the full picture. Political scientist Nina Tannenwald at Brown University has documented at least four moments when U.S. leaders seriously considered using nuclear weapons, during the Korean War, the Vietnam War, and the 1991 Gulf War, and chose not to. In each case, deterrence alone didn’t explain the decision. What also mattered was a growing moral and political inhibition, a sense that these weapons occupied a category apart from all other military tools.
This taboo has been reinforced over decades by international treaties, public opinion, and the shared understanding that crossing the nuclear threshold would fundamentally change the world’s relationship with warfare. It’s not a law of physics. It’s a norm, and norms can erode.
The Current State of Global Arsenals
Nine countries possess nuclear weapons, and collectively they hold an estimated 12,241 warheads as of January 2025, according to the Stockholm International Peace Research Institute. About 3,912 of those are actively deployed on missiles or at bases with operational forces. The rest sit in central storage but could be readied for use.
The United States and Russia dominate the count, with military stockpiles of roughly 5,328 and 5,580 warheads respectively. China’s stockpile has grown to about 500 warheads, a significant expansion from its historically modest arsenal. India holds around 180, Pakistan 170, and North Korea an estimated 50. The United Kingdom, France, and Israel round out the list.
More concerning than the raw numbers is the trajectory. Countries are modernizing their arsenals and expanding the role nuclear weapons play in their military planning. The Bulletin of the Atomic Scientists noted in 2025 that nuclear-armed states are investing hundreds of billions of dollars in weapons capable of destroying civilization, and that countries without nuclear weapons are openly considering developing them for the first time in decades.
Arms Control Is Falling Apart
The architecture that kept nuclear competition within agreed limits is crumbling. The New START treaty between the United States and Russia, which caps deployed strategic warheads, was extended through February 4, 2026, but Russia suspended its participation in 2023, and no replacement negotiations are underway. When it expires, there will be no binding agreement limiting the two largest nuclear arsenals for the first time since 1972.
Other pillars have already fallen. The Intermediate-Range Nuclear Forces Treaty ended in 2019. The Open Skies Treaty, which allowed surveillance flights to build mutual confidence, collapsed in 2020-2021. High-level contacts among nuclear powers are, in the words of the Bulletin, “totally inadequate given the danger at hand.” Without these frameworks, each side has less visibility into what the other is doing, and less reason to exercise restraint.
Where the Risks Are Highest
Several active conflicts involve nuclear-armed states or their close allies, and each carries some probability of escalation.
The war in Ukraine is the most frequently cited risk. Russia’s nuclear doctrine contains deliberate ambiguities about when escalation would be justified. Analysis of Russian strategic documents shows that since 2010, Russia has actually raised its nuclear threshold compared to the 1990s, relying more on conventional forces and information warfare. But critical questions remain unanswered: Would a military defeat that threatened the survival of the Russian government trigger nuclear use? Would a successful Ukrainian attack on Crimea cross that line? Russia’s own doctrine offers fuzzy definitions of what constitutes an “existential threat,” which means even Russian leaders may not know exactly where the line is until they’re standing on it.
South Asia presents a different kind of danger. In 2025, India and Pakistan experienced their most serious military confrontation in years, involving airstrikes, drones, cyber attacks, and naval maneuvers. A U.S.-brokered ceasefire paused the fighting in May, but cross-border violations continue. India maintains a No First Use policy paired with a promise of massive retaliation. Pakistan, which lacks India’s conventional military advantage, pursues what it calls “full-spectrum deterrence,” meaning it reserves the right to use nuclear weapons at any level of conflict to offset India’s larger army. That asymmetry creates a narrow space for miscalculation during a crisis.
The Middle East adds another layer. The Bulletin warned that conflict in the region “threatens to spiral out of control into a wider war without warning,” in a neighborhood where Israel possesses an estimated 90 warheads and Iran’s nuclear program remains a persistent flashpoint.
Tactical Nuclear Weapons Lower the Threshold
Not all nuclear weapons are city-destroying strategic warheads. Tactical nuclear weapons are designed for battlefield use, with yields ranging from fractions of a kiloton to about 50 kilotons. For comparison, the bomb dropped on Hiroshima was 15 kilotons. Some tactical weapons are smaller than that; others are several times more powerful. Their delivery systems typically have ranges under 310 miles, designed for regional conflict rather than intercontinental strikes.
The existence of these smaller weapons creates a dangerous logic. Because they’re less destructive than strategic warheads, some military planners treat them as usable, a way to signal resolve or break a battlefield stalemate without triggering civilization-ending retaliation. Russia, Pakistan, and the United States all maintain tactical nuclear arsenals. The concern is that using even one of these weapons, however “limited” the blast, would shatter the 80-year taboo and make further use far easier to justify.
New Technology Compresses Decision Time
Two technological developments are making the nuclear landscape less stable: hypersonic missiles and artificial intelligence.
Hypersonic weapons travel at five or more times the speed of sound on flatter, more maneuverable trajectories than traditional ballistic missiles. Their intended target is nearly impossible to determine until the final seconds of flight, and a defending country can’t tell whether an incoming hypersonic missile carries a conventional or nuclear warhead. This “warhead ambiguity” can heighten tensions during a crisis, because a leader seeing an incoming missile has to decide how to respond without knowing what it’s carrying. The time reduction compared to ballistic missiles is only a few minutes, and analysts at Air University note that this doesn’t fundamentally undermine the U.S. second-strike capability, since the whole point of second-strike forces is that they survive an initial attack. But for countries with smaller or more vulnerable arsenals, those lost minutes could matter enormously.
Artificial intelligence poses a subtler risk. No nuclear-armed state has formally handed launch authority to a machine, and there is general agreement among these countries that nuclear decisions should not be delegated to AI. The danger is more gradual. As AI tools become integrated into military early warning and decision support systems, research from West Point’s Modern War Institute suggests that human operators tend to defer to AI recommendations, even when other information suggests they shouldn’t. This is known as automation bias, and it gets worse, not better, as AI systems demonstrate competence in complex scenarios like wargames. Once defense planners start viewing AI-generated recommendations as equal or superior to human judgment, the practical distinction between “AI advises” and “AI decides” begins to blur. In a crisis where decisions must be made in minutes, an AI system flagging an incoming attack could push human operators toward responses they might not otherwise choose.
What Keeps the Taboo Intact
Despite all of these pressures, powerful forces still work against nuclear use. Deterrence remains functional among the major powers: any nuclear strike on the United States, Russia, or China would invite devastating retaliation, making it an act of national suicide. The global norm against nuclear weapons, while under strain, still carries real political weight. A country that used nuclear weapons would face immediate and severe isolation from the international community, economic consequences that would dwarf any existing sanctions regime, and a permanent place in history alongside the destruction of Hiroshima.
The most likely path to nuclear use isn’t a calculated, cold-blooded decision by a rational leader. It’s a scenario where a government feels cornered, where conventional military options have failed, where ambiguous intelligence and compressed timelines force a decision before anyone fully understands what’s happening. The risks are not evenly distributed across time. They spike during crises, and the world currently has several crises running simultaneously, with fewer diplomatic channels and arms control agreements to manage them than at any point in decades.
The honest answer to the question is that the probability of nuclear weapons being used again is not zero, and it’s higher than it was ten or twenty years ago. It remains unlikely in any given year, but “unlikely” across enough years of elevated risk eventually becomes “likely enough to demand attention.” The tools that kept the world safe during the Cold War, deterrence, arms control, direct communication between adversaries, are all weaker than they used to be, and nothing new has replaced them.