Nuclear Weapons
The nine arsenals, the quiet modernization wave, the collapse of arms control, and why the deterrence framework that held for eighty years is more fragile than it looks.
(Federation of American Scientists / SIPRI estimates; ~9,500 in active military stockpiles)
(roughly 5,400 US and 5,500 Russian; the strategic-balance core)
(Israel does not officially confirm; the others have confirmed at various points)
A note on framing. Nuclear weapons largely vanished from public attention after the Cold War. They are coming back. The combination of the Russia-NATO confrontation, Chinese arsenal expansion, the collapse of nearly every major arms-control treaty, and the technological pressures of hypersonics, AI integration, and missile defence is producing a strategic environment that has not existed before. This page tries to walk through the picture as it actually is in 2026, not as people who grew up after the Cold War were taught it would stay.
Who has them and how many
The nine nuclear-weapon states are not a uniform group. The differences in arsenal size, delivery systems, doctrine, and political situation matter for understanding the strategic picture.
United States. Roughly 5,400 total warheads, with about 1,700 deployed on strategic delivery systems (intercontinental ballistic missiles, submarine-launched ballistic missiles, bombers) under the New START framework that expires in February 2026. The US is undertaking the most expensive nuclear modernisation since the early Cold War: new Sentinel ICBMs replacing Minuteman III, Columbia-class submarines replacing Ohio-class, B-21 Raider bombers entering service, new W93 warhead in development, and modernised cruise missiles. Total program cost estimated at over $1.5 trillion across decades.
Russia. Roughly 5,500 total warheads, with about 1,500 strategic deployed. Modernisation has been substantial: new RS-28 Sarmat heavy ICBMs partially deployed, new Borei-class strategic submarines in service, new Tu-160M2 bombers in production, hypersonic Avangard glide vehicles, and the controversial Burevestnik nuclear-powered cruise missile and Poseidon nuclear-armed underwater drone. Russia has also expanded its sub-strategic (tactical) nuclear capability, which it has explicitly threatened to use during the Ukraine war.
China. The most rapidly expanding arsenal. Roughly 600 warheads in 2024, projected by US Department of Defense estimates to reach over 1,000 by 2030. New ICBM silo fields documented by satellite imagery. New Type 096 strategic submarines in development. The DF-41 ICBM and DF-26 intermediate-range missile (capable of conventional or nuclear) substantially deployed. Chinese doctrine is officially "no first use," though the practical implications are debated. The expansion is the largest peacetime nuclear build-up since the early Cold War.
France. Roughly 290 warheads, on submarine-launched and air-launched delivery systems. Doctrine emphasises strategic autonomy. Modernisation is steady: new SNLE 3G submarines and ASN4G air-launched cruise missile in development.
United Kingdom. Roughly 225 warheads, on submarine-launched delivery only. The UK announced a partial reversal of post-Cold War warhead reductions in 2021, raising the cap from 180 to 260. Dreadnought-class submarines are replacing Vanguard-class. The UK depends on the US for some warhead components and missile leasing.
India. Roughly 170 warheads, on a mix of land, sea, and air delivery. Doctrine is "no first use against non-nuclear states" and "credible minimum deterrence," with some evolution toward more flexibility. New Arihant-class strategic submarines and Agni-V/VI ICBMs extending range. India's modernisation is steady but lower-tempo than the major powers.
Pakistan. Roughly 170 warheads. Doctrine emphasises tactical nuclear weapons to deter Indian conventional superiority. Continued production of fissile material, new short-range systems (Nasr) intended for battlefield use, and growing concerns - particularly in Western and Indian analysis - about command-and-control during a crisis. Western assessments often rate Pakistan as among the more dangerous of the nine because of doctrine, political instability, and the possibility of fissile material or weapons falling outside state control. Pakistani officials reject this characterisation, citing the Strategic Plans Division's command-and-control architecture and a strong service record on physical security; the empirical case for either reading rests on outside intelligence that is not public.
Israel. Estimated at roughly 90 warheads, with delivery via aircraft, submarines (Dolphin-class equipped with cruise missiles), and Jericho ballistic missiles. Doctrine is "amimut" (deliberate ambiguity) - neither confirmed nor denied. Modernisation is opaque but presumed continuing.
North Korea. Estimated at roughly 50 warheads, with growing fissile-material production and increasingly capable delivery systems. Recent tests have demonstrated tactical nuclear systems, hypersonic claims, and submarine-launched capability of debated reliability. Doctrine has shifted toward more explicit threats of first use under broader conditions.
The collapse of arms control
From 1969 to about 2010, a network of arms-control treaties produced sustained reductions in nuclear arsenals and increased transparency. Most of that network has now collapsed.
Anti-Ballistic Missile Treaty (1972, US-Soviet). Limited missile-defence systems to preserve mutual vulnerability and stability. The US withdrew in 2002 to develop missile defence against rogue states. Russia argued this destabilised strategic balance. The argument shaped subsequent Russian investments in delivery systems designed to penetrate or evade defences.
Intermediate-Range Nuclear Forces Treaty (1987, US-Soviet then US-Russian). Eliminated all ground-launched missiles with ranges between 500 and 5,500 kilometres. The US withdrew in 2019 citing Russian violations (the 9M729 ground-launched cruise missile in particular). Russia denied the violation, asserted the missile was within treaty parameters, and counter-claimed that the US Aegis Ashore deployments in Romania and Poland - with Mk-41 launchers technically capable of firing Tomahawk cruise missiles - were themselves INF violations. Russia formally suspended in response to the US withdrawal. Both sides have since deployed previously prohibited systems. China was never a party and has built extensive intermediate-range capability throughout the treaty period.
Open Skies Treaty (2002, multilateral). Allowed reciprocal aerial reconnaissance flights for confidence-building. The US withdrew in 2020. Russia suspended participation in 2021. The treaty is effectively dead for the original purpose.
New START (2010, US-Russian). Limits each side to 1,550 deployed strategic warheads on 700 delivery systems. Russia "suspended" participation in February 2023 (without formal withdrawal), citing US support for Ukrainian military operations - including alleged drone attacks on Russian strategic-bomber bases that Russia argued violated the spirit of the treaty - and broader Western "hostile policy" as grounds. The US disputed this characterisation and continued partial compliance unilaterally. The treaty expires in February 2026 with no agreed replacement. As of 2026, this leaves the US and Russia for the first time since the 1970s without formal limits on their strategic arsenals.
Comprehensive Test Ban Treaty (1996, multilateral). Bans all nuclear testing. Has been signed by most states but not formally entered into force because key signatories have not ratified - the US itself has never ratified despite signing in 1996, and China, India, Pakistan, North Korea, and several others are in similar positions. Russia withdrew its ratification in 2023, partly framed as response to continued US non-ratification. There has been no nuclear testing by major powers since 1996, but the formal restraint has weakened.
Treaty on the Non-Proliferation of Nuclear Weapons (1968, multilateral). The foundational global treaty restricting the spread of nuclear weapons to states that did not have them in 1967. Generally considered intact but stressed: India, Pakistan, Israel never joined; North Korea withdrew; concerns about Iran, Saudi Arabia, possibly others considering proliferation.
What remains: the NPT (stressed), some bilateral US-Russian incident-prevention measures, and the unwritten norm against nuclear use that has held for 80 years. The architecture that constrained how many weapons existed and how transparently they were managed has substantially come apart over the past 25 years.
The new instability factors
Several technological and geopolitical factors are changing the strategic environment in ways the existing deterrence framework was not designed for.
Three-way arsenal dynamics. Cold War deterrence theory was a two-player problem. The current US-Russia-China three-way is structurally different: any two-way arms-control agreement between the US and either Russia or China provides incentive for the third party to grow toward parity. Negotiating a stable three-way limit is much harder. The US is, for the first time, planning for a possible deterrence requirement against two near-peer adversaries simultaneously.
Hypersonic weapons. Hypersonic glide vehicles and cruise missiles travel at five-plus times the speed of sound and can manoeuvre in flight, complicating missile defence. Russia, China, and the US have all deployed or are deploying hypersonic systems. Some are nuclear-capable; some are conventional. The blurring between conventional and nuclear delivery systems is itself a stability concern.
Tactical/sub-strategic nuclear weapons. Lower-yield nuclear weapons designed for battlefield use have proliferated in Russian, Pakistani, and increasingly Chinese arsenals. The US has reintroduced lower-yield warheads (W76-2) on submarine-launched missiles. The argument for tactical weapons is that they provide options short of strategic exchange. The argument against is that they lower the threshold for any nuclear use, making escalation more likely.
AI and command-and-control. Decision-support systems, automated detection, and AI integration into command-and-control are creating new questions about reaction times, automation bias, and the human role in nuclear decisions. The US, Russia, and China have all stated commitment to "human in the loop" for nuclear-use decisions, but the practical implications under crisis time pressure are debated.
Cyber threats to nuclear systems. The vulnerability of command-and-control, early-warning, and weapons systems to cyber attack is now a recognised concern. Several near-misses in early-warning systems across decades suggest the consequences of compromise could be severe. Defensive investment is substantial; whether it is adequate is opaque by design.
The Ukraine precedent. Russia's repeated nuclear threats during the Ukraine war, the detailed Russian doctrine modifications, and the public discussion of conditions for nuclear use have normalised explicit nuclear coercion in ways the post-Cold War period had largely avoided. Some context-setting matters here: nuclear signalling has been a feature of Western strategic communication too (the 2017-18 "fire and fury" exchange with North Korea, US "calculated ambiguity" language about Iran, occasional French and UK statements about extended deterrence), so the pattern is one of all major nuclear powers becoming more comfortable with explicit nuclear rhetoric, with Russia going furthest. Whether this generalises to other conflicts is one of the most consequential open questions.
Proliferation pressure
The number of nuclear-weapon states has been remarkably stable for decades. South Africa dismantled its arsenal in 1991. Libya gave up its programme in 2003. Iran has been the long-running case of programme that may or may not have crossed the threshold. North Korea did cross. The current pressure is broader.
Iran. The 2015 nuclear deal substantially constrained the Iranian programme; the US withdrawal in 2018 began its unwinding; Iranian capability has progressed substantially since 2020. As of 2026, IAEA inspection reports document weapons-grade-adjacent enriched-uranium stockpiles sufficient for several weapons-worth of fissile material; US intelligence community assessments and analyses from the Institute for Science and International Security and the Arms Control Association put the breakout time at weeks rather than months, with weaponisation (assembling a deliverable weapon from enriched material) requiring additional months. Iranian officials maintain that the programme remains civilian and that breakout-time estimates conflate stockpile size with weapons intent. Whether and when Iran crosses the threshold is one of the highest-stakes questions in current international politics. Israeli action against the Iranian programme is a recurring possibility.
Saudi Arabia. Saudi officials have repeatedly stated that if Iran develops a weapon, Saudi Arabia will pursue equivalent capability, presumably through a Pakistani arrangement or rapid indigenous development. The credibility of these statements is debated; the financial and technical pathway is plausible.
South Korea. Polling shows substantial public support for an indigenous nuclear weapon. Senior officials have raised the question publicly. The North Korean threat and uncertain US security commitment under different administrations create real pressure. South Korean technical capability would allow rapid development if political decision were made.
Japan. Officially committed to non-nuclear status, but with substantial latent capability (extensive civilian plutonium stocks, advanced delivery technology). Periodic public debate about whether the security environment justifies reconsideration. Currently low probability of change, but the underlying capability is real.
Other possibilities. Turkey under different leadership scenarios, Egypt under acute regional pressure, Poland or other European states under collapsed-NATO scenarios. None of these is currently active; each is more thinkable than at any point since the 1970s.
The non-proliferation regime is held together by a combination of treaty obligations under the NPT, security guarantees from existing nuclear-weapon states (especially US extended deterrence over NATO allies, Japan, South Korea, and Australia), and economic and political costs that the US, EU, and UN Security Council have applied to states pursuing nuclear weapons outside the NPT framework. The regime reflects a 1968 settlement that allowed the five then-existing nuclear powers (US, USSR/Russia, UK, France, China) to keep their arsenals while restricting others. There are no agreed objective criteria for which states should have nuclear weapons - every state that has them wanted them for the same reason every state without them might want them (deterrence against being attacked or coerced). The practical concern that has held the regime together is narrower: wider proliferation increases the risks of accidental use, theft, regional escalation dynamics that none of the participating states can fully control, and the possibility of weapons or fissile material reaching governments or non-state actors whose restraint cannot be relied on. NPT supporters argue this concern has been validated by the relatively small number of crossings since 1968 and by the costs already evident from the existing nine-state arsenal. Critics argue the same restraint logic should apply more demandingly to the existing nuclear powers, several of which have engaged in nuclear coercion, modernised their arsenals rather than reduced them, and behaved in ways that test the assumption of responsible stewardship. Each of these elements has weakened somewhat over the past decade. Whether the regime holds depends on whether high-profile proliferation occurs and what the international response is.
The paths from here
Continued steady-state arms racing without a major incident
Modernisation continues across all nine arsenals. New START expires without replacement. Tactical nuclear capabilities expand. No nuclear use; no major proliferation event. The total warhead count slowly rises after decades of decline. This is roughly the current trajectory.
Crisis without nuclear use produces wake-up effect
A serious nuclear crisis - close call in the Ukraine war, Taiwan-related confrontation, India-Pakistan escalation - produces sustained political attention to arms control without actually crossing the use threshold. Limited new agreements emerge. The sense that the system was working has been quietly false; the wake-up reactivates serious diplomacy.
Tactical use in a regional conflict
A nuclear-armed state uses one or a few low-yield weapons in a regional conflict. The eighty-year nuclear taboo breaks. International response is severe but not catastrophic. The world adjusts to a new normal where nuclear use is rare but not impossible. This is the scenario nuclear strategists have studied most and is genuinely contested in probability.
Cascade proliferation
Iranian nuclear capability triggers Saudi acquisition. South Korea moves toward indigenous capability under acute North Korean pressure or US-alliance uncertainty. Other states follow. The non-proliferation regime substantially erodes. Strategic stability becomes much harder to maintain across more arsenals.
Renewed arms control on a trilateral basis
The US, Russia, and China negotiate some form of trilateral framework, even imperfect, providing limits and transparency. China's resistance to formal arms control eases as its arsenal approaches parity. The framework constrains the next phase of build-up. This is the optimist case; it requires diplomatic and political conditions that have not been present recently.
A catastrophic incident reshapes everything
Nuclear weapons are used in a strategic exchange or a major terrorist scenario. The casualties are large enough that the post-incident political environment fundamentally changes. The post-1945 nuclear era ends. What replaces it is unpredictable and could be either much more restrictive or much more chaotic.
Where serious analysts disagree
The deterrence framework is more fragile than the public realises
The combination of three-way dynamics, modernisation, doctrinal expansion, and arms-control collapse has produced a strategic environment less stable than at most points during the Cold War. The eighty-year non-use record is partly luck and partly structure; the structural component has weakened. Sustained attention is needed before a crisis tests the system.
Held by: Vipin Narang (MIT), Pavel Podvig (Geneva-based Russian-strategic analyst, runs russianforces.org, operates outside the Russian state system but draws heavily on Russian open-source material), the Federation of American Scientists' nuclear team, Anne Neuberger, and a substantial part of the senior nuclear-policy community. Their case has been strengthened by the Russian doctrinal moves and Chinese build-up.
Deterrence is robust; alarm is overstated
The eighty-year non-use record is not luck. The fundamental logic of mutual destruction continues to hold. Modernisation is replacement of aging systems, not arsenal expansion. The arms-control collapse is partly verifiable continuity through tacit cooperation. The system is more stable than the alarmed reading conveys.
Held by: parts of the realist IR community, Kenneth Waltz's intellectual successors (most prominent in the academy), some retired senior US strategic-planning officials. The case is supported by the actual non-use record.
China's build-up is the central new variable
The most consequential change in the next five years is the Chinese arsenal moving toward parity with the US and Russian arsenals. The nature of three-way deterrence has not been worked through carefully in either US or Chinese doctrine. The risk is that a Taiwan-related crisis would unfold under conditions where the parties' theories of deterrence are less developed than during the Cold War.
Held by: Brad Roberts (Lawrence Livermore National Laboratory; former US Deputy Assistant Secretary of Defense for Nuclear and Missile Defense Policy), Christopher Twomey (Naval Postgraduate School), parts of US Strategic Command published analysis, and a small set of Chinese-origin scholars working at Western institutions (Tong Zhao at Carnegie, Fiona Cunningham at University of Pennsylvania) who track Chinese nuclear doctrine from inside the relevant linguistic and intellectual context. Their case is supported by the documented Chinese arsenal trajectory; counter-readings from PLA-affiliated authors argue the build-up is a defensive response to US missile defence and counterforce capabilities rather than a move toward parity-seeking.
Tactical nuclear weapons are the most dangerous category
Lower-yield "useable" nuclear weapons in expanding arsenals (Russia, Pakistan, US W76-2, Chinese systems) lower the threshold for any nuclear use. Once one tactical nuclear weapon has been used, the path to escalation is poorly mapped. The doctrinal embrace of tactical weapons is a significant shift from the Cold War view that all nuclear use was strategic.
Held by: Hans Kristensen (FAS), Bonnie Jenkins (former US Under Secretary of State), and a strand of arms-control research. Their case has been strengthened by Russian doctrine evolution.
Nuclear abolition is the only real solution
The cumulative risk of nuclear use, accidents, theft, and proliferation across the long term is high enough that maintaining the system indefinitely is reckless. The Treaty on the Prohibition of Nuclear Weapons (entered force 2021, not joined by any nuclear-weapon state) represents the right direction. Realist arguments about deterrence underestimate long-run risk in favour of short-run stability.
Held by: ICAN (International Campaign to Abolish Nuclear Weapons), the coalition of mostly non-Western middle powers (Austria, Ireland, Mexico, South Africa, Indonesia, Malaysia, New Zealand, the Holy See, and a long list of Global South states) that drove the TPNW through the UN over the opposition of all nine nuclear-weapon states and most of NATO, and parts of the disarmament community. Several former senior US officials have advocated similar goals through different means - Sam Nunn, William Perry, George Shultz, and Henry Kissinger were the prominent "Four Horsemen" group calling for major US-Russian reductions. The case is morally serious; politically it is marginal in nuclear-armed states but mainstream in most of the Global South.
None of these readings is fully right or wrong. What can be said from the available evidence: the deterrence system has held for eighty years through a mix of structure and luck; the structural component has weakened over the past two decades through arms-control collapse, modernisation, doctrinal expansion, and three-way dynamics that the existing framework was not designed for; tactical nuclear weapons and the proliferation pressure are particularly concerning new variables; and the political attention to nuclear questions is now far below what the technical situation warrants.
What this means for you
If you read foreign-policy coverage
Nuclear questions are systematically under-covered relative to their importance. When you encounter coverage of Ukraine, Taiwan, the Middle East, India-Pakistan, or Korean peninsula tensions, the nuclear dimension is often the most consequential layer and often the least discussed. Reading specialised sources (Bulletin of the Atomic Scientists, FAS Nuclear Notebook, Arms Control Today, Russia Matters) alongside general coverage helps.
If you live in a likely target area
Major cities, major military installations, and missile-defence sites are the most likely targets in any large-scale exchange. Personal preparation cannot meaningfully change outcomes for nearby strikes; modest practical preparation (basic supplies, awareness of shelter options) can affect outcomes for fallout from more distant strikes. The risk of any specific person being affected in any given year remains low; sustained attention to keeping it that way is reasonable civic engagement.
If you vote on national-security policy
The questions that matter most for nuclear stability are unsexy and not politically salient: arms-control negotiating positions, nuclear-modernisation specifics, alliance commitments and extended deterrence, command-and-control investments. Politicians who engage these substantively are doing some of the most important work; politicians who treat nuclear weapons as ordinary campaign material usually have not done the work.
If you have technical training
The arms-control verification community, the nuclear-stability academic community, and government nuclear-policy positions are persistently understaffed relative to the importance of the work. Career paths exist in technical verification, policy analysis, intelligence, and diplomacy. The community has tried to recruit more young technical talent for years; the supply has been thin.
If you support arms-control work
Specific organisations have outsized impact relative to their size: the Bulletin of the Atomic Scientists, the Federation of American Scientists, the Arms Control Association, the Nuclear Threat Initiative, and the academic community at MIT, Stanford, Stimson Center, and elsewhere. Direct support is high-leverage; the nuclear-policy community is small enough that incremental funding meaningfully changes what work gets done.
