Nuclear Proliferation as a Contemporary Global Concern

Nuclear proliferation is the one contemporary global concern in which a single failure is not recoverable. Climate damage accumulates, rights abuses can be reversed, democracies rebuilt — a nuclear detonation ends the argument. The concern is therefore not only how many states hold the bomb, but the bargain that persuades most states never to seek it, the machinery that verifies restraint, and whether the possessors are keeping their side of a promise made in 1968.

Fixing the Vocabulary

• Proliferation is the spread of nuclear weapons, fissile material and delivery technology — and the word hides two different problems.
  • Horizontal proliferation is growth in the number of states holding weapons, which the regime was built to stop.
  • Vertical proliferation is the growth and modernisation of arsenals by states already inside the club.
  • The regime’s asymmetry is that it polices the horizontal and barely touches the vertical — the grievance organising almost every Southern critique.
• Latency, or the nuclear hedge, is holding the full fuel cycle and the engineering competence without assembling a weapon.
  • Japan is the standard illustration: large separated plutonium holdings, enrichment capacity, advanced launchers, a declared non-nuclear policy.
  • Latency is lawful because Article IV protects the “inalienable right” to peaceful nuclear technology — the treaty’s structural loophole.
  • Nuclear reversal is the opposite movement, achieved by South Africa, Ukraine, Belarus, Kazakhstan, Libya, Brazil, Argentina, Sweden and Taiwan.
• Dual use makes the problem intractable: the cascade that fuels a reactor at 3–5% enrichment yields weapons-grade uranium at 90% with more machines running longer.
  • There is no technical line between civil and military programmes, only a line drawn by intent, and intent cannot be inspected.
  • The physical threshold is fissile material — roughly 15–25 kg of highly enriched uranium or 5–8 kg of plutonium for a first device.
  • Three routes reach it: dedicated military facilities, diversion from a civil programme, or theft and black-market purchase.

Why it is a global and not a regional concern

• Consequences do not stay local. Fallout, climatic effects and the collapse of the non-use tradition reach states with no part in the quarrel; modelling of even a limited regional exchange projects agricultural collapse far beyond the belligerents.
• Acquisition is contagious. Each entrant alters neighbours’ calculations — the Soviet bomb after the American, the Chinese after the Soviet, the Indian after the Chinese, the Pakistani after the Indian.
• Supply networks are transnational. The A.Q. Khan network moved centrifuge designs between Pakistan, Libya, Iran and North Korea through private brokers in a dozen jurisdictions.
• Non-state actors cannot be deterred, because deterrence needs a return address — which is why fissile-material security became a separate agenda after 2001.
• The regime is a public good with a free-rider structure: all benefit from others’ restraint, each has reason to hedge, and those best able to enforce restraint are least willing to accept limits themselves.

The regime asks the many to accept permanent restraint in exchange for a promise of eventual disarmament by the few — and the few have not delivered.

Supply-side and demand-side models

Dimension	Supply-side model	Demand-side model
Assumption	Proliferation follows available capability	Proliferation follows desire; remove the motive and capability is irrelevant
Instruments	Export controls, safeguards, trigger lists, interdiction, denial of enrichment and reprocessing	Security guarantees, alliances, extended deterrence, regional settlements, weapon-free zones
Support	Slowed Iraq, Libya and Iran; lengthened lead times everywhere	Japan, Germany, South Korea and Australia refrained because they were protected
Failure mode	Determined states build indigenously — India, Pakistan, North Korea	Guarantees are only as good as the guarantor; alliance doubt is itself a driver
Political cost	Reads as technological apartheid to recipients	Requires great powers to underwrite others’ security open-endedly

• The two are complements, not rivals: supply control buys time, and only demand management converts that time into permanent restraint.

Why States Build the Bomb, and Why Most Do Not

Scott Sagan’s three models

• Scott Sagan‘s framework is the field’s most useful organising device, because it refuses the realist assumption that proliferation is always about security and yields a different remedy for each pathway.

Model	Driver	The bomb is	Cases	Remedy
Security	External threat under anarchy	A response to threat	USSR after 1945; China after the Taiwan Strait crises; India after 1962 and 1964; Pakistan after 1971	Credible guarantees or removal of the threat
Domestic politics	Scientists, weapons laboratories, the military-industrial complex, politicians using the issue	A solution seeking a problem	The French scientific lobby; atomic-energy establishments as autonomous actors; South Africa’s military bureaucracy	Civilian control, budget transparency, breaking laboratory monopolies on information
Norms and symbols	Identity and prestige — the bomb as a marker of modernity and standing	A normative symbol	De Gaulle’s France; the reading that Security Council seats and warheads are one list; Ukraine’s renunciation as a bid for European identity	Change what confers status — stigmatise possession, reward restraint

• Sagan’s decisive move is that the security model is over-applied, because it is the story states tell afterwards, while the decision record often shows domestic entrepreneurs and identity claims doing the work.

Hymans, Solingen and the technological account

• Jacques Hymans locates the choice in the leader’s national identity conception, and identifies the oppositional nationalist — who sees a rival as both threatening and inferior — as the type that reaches for the bomb.
  • This explains the puzzle of timing: capability often precedes decision by decades, and what changes is who is in office.
• Etel Solingen explains it by domestic political economy and the ruling coalition’s survival strategy.
  • Outward-oriented, export-led regimes forgo nuclear weapons because markets, investment and technology are worth more than the bomb.
  • Inward-looking, nationalist regimes under sanctions find that nuclear programmes reinforce their model of rule.
  • The prediction has largely held — liberalising economies denuclearise — which is why sanctions sometimes work by strengthening internationalists and sometimes fail by destroying them.
• The older technological-determinist account, that capability produces weapons automatically, is refuted decisively: dozens of states could build the bomb and have not.

The incentives and the disincentives

Incentives to acquire	Disincentives
Security against an existential threat — Israel after 1948, North Korea against forward-deployed forces	Alliance substitutes — Japan, South Korea, Australia and NATO members under a credible umbrella
Prestige and status, and the belief that Council permanence and nuclear status are one list	Cost and technical difficulty — fuel cycle, weaponisation, delivery, testing, command
Strategic autonomy from an ally’s decisions — the British and, more sharply, the French programme after Suez	Normative cost — sanctions, technology denial and diplomatic isolation attach to acquisition
Alliance doubt that the umbrella will open — the anxiety now visible in Seoul, Tokyo, Warsaw and Riyadh	Security-dilemma effects — acquisition can invite pre-emption, as at Osirak in 1981 and Al-Kibar in 2007
Domestic political pressure, paying a legitimacy dividend in democracies and autocracies alike	Leadership decisions — Libya dismantled in 2003 to normalise relations, though the regime’s later fall taught the opposite lesson
Bargaining leverage — capability traded for sanctions relief, guarantees or recognition	Trade and investment consequences, which moved Brazil and Argentina to mutual inspection
A weapon of last resort against a defeat that would end the state	Domestic opposition, though this cuts both ways where majorities favour a bomb

Deterrence: Making the Weapons Work Without Using Them

• Deterrence is dissuasion by threat — persuading an adversary that costs will exceed benefits, so the act is never attempted.
  • It is a psychological relationship, existing in the adversary’s mind, which is why its failures are failures of perception.
  • Its three requirements are capability, credibility and communication.
  • Punishment threatens retaliation; denial threatens to make the objective unattainable. Immediate deterrence addresses a crisis, general deterrence operates continuously.
  • Central deterrence protects one’s own territory; extended deterrence protects an ally and is far harder to make credible.
• The extended-deterrence credibility problem is the pivot of alliance politics and of today’s cascade anxiety: the ally asks whether Washington would trade New York for Seoul, and no declaration answers it.
  • The historical solutions — forward deployment, nuclear sharing, tripwire forces — all work by removing the option of not honouring the threat.

The doctrinal chain and the technical architecture

• Massive retaliation (1954) threatened overwhelming response at times and places of American choosing; its flaw was that it offered a president only capitulation or holocaust.
• Flexible response added conventional and limited nuclear rungs, restoring credibility at the cost of making use imaginable.
• Mutual assured destruction obtained once both sides held survivable forces, and is a condition, not a policy: each can absorb a first strike and still destroy the other, so starting a war is suicide.
  • It rests wholly on second-strike survivability, which is why the triad — land-based missiles, ballistic-missile submarines, bombers — matters, the submarine leg being the survivable one.
• Counterforce targets weapons and command; countervalue targets cities and industry.
  • The paradox is that counterforce is more usable and therefore more destabilising, inviting a “use them or lose them” calculation, while countervalue is more horrifying and more stabilising.
• Albert Wohlstetter‘s “delicate balance of terror” denies that deterrence is automatic: retaliatory forces must survive surprise attack, survive to reach targets and survive defences, none of it guaranteed.
• Thomas Schelling supplied the ideas that make nuclear bargaining intelligible.
  • The threat that leaves something to chance: a deliberately suicidal response is not credible, so states manufacture credibility by generating risk they do not fully control.
  • Tacit bargaining and focal points: adversaries coordinate on salient thresholds, and complete non-use is the strongest focal point in politics because any breach opens a slope with no next stopping place.
• Bernard Brodie stated the founding proposition — the military’s purpose in the nuclear age is not to win wars but to avert them; Herman Kahn insisted that working through escalation ladders was necessary precisely to keep deterrence credible.

Security dilemma, spiral and deterrence models

• The security dilemma underlies every arms race: defensive measures are indistinguishable from preparations for attack, so they reduce others’ security and provoke matching effort.

	Deterrence model	Spiral model
The adversary is	Aggressive and opportunistic, restrained only by strength	Insecure and reactive, hostile because frightened
War is caused by	Weakness and appeasement	Mutual fear and misperception
Response	Firmness, armament, alliance, visible resolve	Reassurance, transparency, arms control
Analogy invoked	Munich 1938	July 1914
Risk if wrong	You appease an aggressor	You create the enemy you feared

• Robert Jervis‘s point is that the choice between the two determines policy, and that an arms race may be a symptom of rivalry rather than its cause — so freezing weapons without settling the quarrel achieves less than it appears.
  • The Cold War race consumed a very large share of Soviet output and contributed to the collapse; it also did not produce war, which is the empirical foundation of nuclear optimism.

The stability–instability paradox

• Glenn Snyder‘s stability–instability paradox holds that stability at the strategic level produces instability below it: if a full exchange is unthinkable, each side is freer to take risks at lower rungs.
• South Asia is where the idea became an operating logic rather than a hypothesis.
  • Nuclear weapons plausibly deterred the large conventional wars of 1965 and 1971, while Kargil occurred in 1999, a year after both states tested, and the decades since have been marked by cross-border terrorism and limited retaliation.
  • India’s search for space beneath the threshold through calibrated conventional response, and Pakistan’s battlefield tactical weapons designed to close that space, are the paradox’s two halves in action.
  • The May 2025 exchange after the Pahalgam attack, in which both states struck each other’s territory before halting within days, showed both that limited war under the nuclear shadow is possible and that the participants stopped quickly.
• The critique is that the paradox assumes escalation can be controlled, and no nuclear dyad has tested that assumption to its limit.

The Sagan–Waltz Debate

• The exchange between Kenneth Waltz and Sagan, conducted across editions of The Spread of Nuclear Weapons: A Debate, is the defining argument of the field, and each side wins a different half.
• Waltz, compressed into “more may be better“, argues that nuclear weapons induce caution and that gradual spread is stabilising.
  • They make the cost of war calculable and catastrophic, removing the miscalculation about relative strength that causes conventional wars.
  • They do not require superiority: a small survivable second-strike force deters a much larger arsenal, so nuclear arms racing is strategically pointless.
  • Deterrence is easy, not delicate — uncertainty about retaliation suffices when the penalty is annihilation.
  • His anchor is that no two nuclear-armed states have fought a full-scale war; nuclear weapons made the Cold War cold, and what John Lewis Gaddis called the long peace among great powers is the outcome to be explained. He extended the claim to South Asia and, controversially, to Iran.
• John Mearsheimer endorses a narrower selective proliferation — stabilising for stable status-quo powers, dangerous for unstable regimes — and argued that Ukraine should have retained a deterrent.
• Sagan does not dispute rational deterrence; he disputes that states are the rational unitary actors it assumes.
  • Arsenals are run by complex military organisations with parochial interests, standard procedures and biases toward offensive doctrine and preventive war.
  • Organisational theory predicts normal accidents: in tightly coupled complex systems, failures are statistically expected rather than aberrant.
  • The record he assembles — false warnings, misread radar and satellite returns, crashed aircraft carrying weapons, unauthorised alerts, near-misses during the Cuban Missile Crisis — shows deterrence nearly failing through machinery rather than choice.
  • Civil–military pathology matters most in new nuclear states, where the military holds custody, oversight is thin and short flight times force hair-trigger readiness.
  • His conclusion is probabilistic: given enough time and enough actors, deterrence will fail somewhere.

Question	Waltz	Sagan
Unit of analysis	The state as rational actor under structure	Organisations inside the state
Is deterrence robust?	Easy and self-enforcing	Delicate, dependent on organisational competence
New nuclear states	Stabilising; entrants learn caution	Destabilising; weak controls, short warning
Near-misses show	The system worked	The system nearly failed, repeatedly, by luck
Policy	Managed gradual spread	Strong non-proliferation, de-alerting, civilian command
Weakness	Small N; no allowance for accident or theft	Cannot explain eight decades without use

• The honest verdict is that Waltz explains the record and Sagan explains the risk: eight decades is real evidence, and a very short run from which to infer safety about a system whose single failure is unbounded.

Norms, Taboos and Epistemic Communities

Deterrence explains why a nuclear-armed state is not attacked. It does not explain why nuclear weapons were never used against non-nuclear adversaries in Korea or Vietnam, why no state has used one in eight decades, or why the regime’s technical rules took the shape they did. That residue is what the constructivist and sociological literature accounts for.

Tannenwald and the nuclear taboo

• Nina Tannenwald argues in The Nuclear Taboo that non-use rests not only on deterrence but on a normative prohibition treating nuclear weapons as categorically unacceptable.
  • A taboo is stronger than a norm: it does not counsel against an act, it makes it unthinkable, so the option is never seriously tabled.
  • Her evidence is the cases deterrence cannot explain — Korea, Dien Bien Phu, Vietnam, the Gulf War — where the adversary could not retaliate and use was still ruled out.
  • The taboo was constructed, not given, emerging from revulsion after Hiroshima and Nagasaki, world opinion, anti-nuclear movements, the diplomacy of non-nuclear states, and accumulating practice.
  • Its effect is constitutive as well as regulative: it defines what kind of state a user would be understood to be, a cost no deterrence calculus captures.
• Her own warning is against complacency: a taboo is a social fact that can erode, and non-use so far is no guarantee. She reads the 2017 prohibition treaty as converting an informal taboo into codified law.
• The critique is that a taboo is hard to distinguish from prudence, since leaders who declined use also had military and escalation reasons. Her reply is that the language of decision shifted over time from tactical calculation to moral impossibility.

T.V. Paul and the tradition of non-use

• T.V. Paul offers a weaker and more durable account, preferring “tradition of non-use” to taboo.
  • A tradition is sustained by reputational and self-interested calculation, reinforced each time it is observed, not by internalised morality.
  • Its mechanism is self-deterrence born of reputational cost: a state using nuclear weapons against a non-nuclear opponent would suffer lasting damage to standing, alliances and legitimacy.
  • The tradition also serves the possessors’ interest in delegitimising acquisition by others — weapons seen as unusable are less worth wanting.
• The difference changes the prediction: a taboo is robust, a tradition is contingent and can be abandoned when the reputational cost seems worth paying.
  • Paul is correspondingly more worried about regional nuclear powers and about tactical weapons, whose smaller yields and battlefield framing are designed to lower the threshold.

Peter Haas and epistemic communities

• Peter Haas defines an epistemic community as a transnational network of professionals with recognised expertise and an authoritative claim to policy-relevant knowledge.
  • Members share causal beliefs, normative commitments, notions of validity and a common policy enterprise.
  • Influence operates through uncertainty: politicians facing a technically opaque problem delegate its framing to those who understand it, and the framing then bounds acceptable policy.
• The arms-control expert community is the paradigm case and shaped the regime at every stage.
  • The civilian strategists — Brodie, Schelling, Wohlstetter, Kahn — invented the vocabulary of deterrence, second strike and stability, and governments adopted their categories wholesale.
  • The same community produced the counter-intuitive proposition that defence is destabilising, which became the ABM Treaty of 1972.
  • The Pugwash Conferences, born of the Russell–Einstein Manifesto, carried ideas across the Iron Curtain when official channels froze, and are credited with contributions to the Partial Test Ban and the ABM Treaty.
  • Verification seismology settled a political dispute technically: proving that underground tests were detectable at distance made a comprehensive ban negotiable, and the International Monitoring System is that community’s monument.
  • Humanitarian-consequences research in the 2010s reframed disarmament away from strategic stability toward catastrophe, and produced the prohibition treaty.
• The limits are equally instructive.
  • Epistemic communities shape means far more than ends: they can say how to verify a test ban, not whether to want one.
  • They are unevenly distributed, so the “expert consensus” the regime encodes has often been the possessors’ consensus.
  • When stakes are high enough, expertise is overridden — the American Senate rejected CTBT ratification in 1999 against near-unanimous technical advice.

Mechanism	Explains	Core claim	Vulnerability
Deterrence	Why nuclear states are not attacked	Rational cost-benefit under threat of retaliation	Silent on non-use against non-nuclear states
Nuclear taboo	Restraint even where use was safe	An internalised prohibition makes use unthinkable	Hard to separate from prudence; norms erode
Tradition of non-use	The same restraint, more thinly	Reputational self-interest, reinforced by repetition	Contingent; tactical weapons built to breach it
Epistemic communities	Why the regime has its technical shape	Expert consensus fills the space uncertainty creates	Shapes means, not ends; concentrated in rich states

• Together they answer what deterrence cannot: the regime rests as much on shared meanings and expert authority as on threats, so its erosion begins in language and legitimacy before it appears in warhead counts.

How the Theoretical Traditions Read the Problem

Tradition	Reading	Prescription
Realism	A rational response to anarchy and threat; the bomb is the ultimate self-help instrument. Waltz sees spread as stabilising, defensive realists fear the security dilemma	Balance, deter, extend guarantees; expect little from law where survival is at stake
Liberal institutionalism	A collective-action problem solvable by regimes that supply information and make cheating detectable	Strengthen the NPT, IAEA and verification; bind states through interdependence
Constructivism	Meaning is socially constructed — British warheads do not threaten Washington and North Korean ones do, which no material theory explains	Change the meaning: stigmatise possession, reward restraint
Marxist and critical theory	An instrument of hierarchy — the core freezing its technological advantage over the periphery, an arms economy serving capital, security defined to exclude the violence of poverty	Expose the centre-periphery interest behind non-proliferation; tie disarmament to economic justice
Copenhagen School	Proliferation is securitised: for Buzan and Wæver security is a speech act, which is why Iranian enrichment is an emergency and Israeli warheads are not discussed	Desecuritise where possible; ask who speaks, about what, to which audience
Feminist IR	Nuclear strategy as gendered discourse — Carol Cohn’s “technostrategic” language sanitising annihilation and coding restraint as weakness	Restore the human consequences the vocabulary erases
Postcolonial critique	The regime as nuclear apartheid, dividing the world permanently by a date	Universal, non-discriminatory, verifiable disarmament

Building the Regime: From the Baruch Plan to the NPT

Monopoly and its collapse, 1945–53

• The regime began with an attempt at international ownership of the atom, and its failure set the pattern for everything since.
  • The Acheson–Lilienthal Report (1946) accepted that the American monopoly would not last and proposed an international authority to own and operate all dangerous nuclear activities, since inspection alone could not work.
  • The Baruch Plan (1946) carried this to the UN Atomic Energy Commission as an International Atomic Development Authority with control, licensing and inspection powers — but attached no veto on enforcement, and American disarmament only after control was in place.
  • The Soviet counter-position was that the United States must disarm first, establishing the sequencing dispute — control before disarmament or the reverse — that has never been resolved.
  • India opposed the Baruch Plan too, on the ground that it prohibited national research and development in atomic energy.
• The Atomic Energy Act of 1946 then legislated the monopoly, vesting title to all fissionable material in the government and classifying all information, even from wartime partners.
  • The result was proliferation by resentment: Britain built its own programme and tested in 1952, and Canada commercialised its expertise instead.
  • The Soviet test of 1949, the British of 1952, the French of 1960 and the Chinese of 1964 each closed a phase. The Chinese test was decisive, making proliferation a global rather than an East–West problem and turning a poor post-colonial state into a nuclear power.

Atoms for Peace and the IAEA

• Eisenhower’s “Atoms for Peace” address (1953) reversed denial, offering technology and material for civil use under international auspices, with contributions of fissile material to an international agency — which became the International Atomic Energy Agency in 1957.
  • The programme delivered rapid diffusion through research reactors, training, declassification and dozens of cooperation agreements, and seeded the proliferation problem it was meant to contain, distributing reactors and trained physicists to India, Pakistan, Israel and Iran.
• The IAEA’s mandate is permanently dual: to accelerate atomic energy’s contribution to peace, health and prosperity, and to ensure assistance is not used militarily.
  • Safeguards rest on four operations — design review, records and accounts, reports, and inspection.
  • Safeguards are not concerned with physical protection of material against theft, which is why a separate nuclear-security architecture had to be built later.

The NPT: the bargain, the pillars, the cut-off

• The Treaty on the Non-Proliferation of Nuclear Weapons opened for signature in 1968, entered into force in 1970, and has 191 states parties — the most widely adhered-to arms-control instrument in existence.
• It rests on a grand bargain of three pillars, and reading it as a bargain rather than a prohibition is the key to every dispute about it.
  • Non-proliferation: Articles I and II bar transfer and acquisition, and Article III requires non-nuclear-weapon states to accept IAEA comprehensive safeguards.
  • Disarmament: Article VI obliges each party to pursue negotiations in good faith on cessation of the arms race and on nuclear disarmament.
  • Peaceful uses: Article IV affirms the inalienable right to nuclear energy for peaceful purposes and to the fullest possible exchange of technology.
• The definitional core is Article IX(3): a nuclear-weapon state is one that manufactured and exploded a device before 1 January 1967.
  • The date captured exactly five states — the United States, the Soviet Union, the United Kingdom, France and China — and no others.
  • It is a calendar, not a criterion. Nothing about it corresponds to responsibility, restraint or capability, and this arbitrariness frozen into law is what India has objected to for six decades.
• Article X permits withdrawal on three months’ notice where “extraordinary events” have jeopardised a party’s supreme interests — the clause North Korea invoked in 2003.

The criticisms

• Structural discrimination: a permanent legal distinction between haves and have-nots created by a date, conferring a status others can never lawfully acquire.
• The disarmament pillar is unenforceable: Article VI requires negotiation in good faith, with no timeline, benchmark or consequence. Five decades on, the five retain thousands of warheads and all are modernising.
• It restrains horizontal but not vertical proliferation — nothing limits the size, quality or deployment of the arsenals it legitimised.
• Asymmetric safeguards: comprehensive and intrusive for the non-nuclear, voluntary offer arrangements for the five, so the inspection burden falls entirely on those who gave up the weapons.
• The stationing loophole: Articles I and II were read as permitting deployment on allies’ territory, which is how NATO nuclear sharing survives inside a non-proliferation treaty.
• The Article IV loophole: the right to peaceful technology includes enrichment and reprocessing, so a state can build the fuel cycle lawfully and then exercise Article X — the break-out pathway.
• Peaceful nuclear explosions: the restrictions were read as one-sided, the five reserving explosive engineering to themselves.
• The security gap: the treaty did nothing about the fears that make states want weapons, and the negative security assurances eventually offered are unilateral, conditional and non-binding.
• No enforcement: non-compliance must be referred by the IAEA Board to the Security Council, where a veto governs the result.

The NPT is not a disarmament treaty and never was; Article VI is a promise to negotiate, and the failure to convert that promise into performance is the regime’s central unresolved grievance.

The extension bargain and the review record

• The NPT was written with a twenty-five-year term and a decision point built in. At the 1995 Review and Extension Conference it was extended indefinitely and without a vote.
  • Extension was purchased with a package: strengthened review, Principles and Objectives including completion of the CTBT and the start of FMCT negotiations, and a resolution on a Middle East zone free of weapons of mass destruction — the price of Arab consent to permanence, and still unimplemented.
• The 2000 Review Conference produced the regime’s high-water mark, the Thirteen Practical Steps, including an “unequivocal undertaking” to accomplish total elimination, early CTBT entry into force, an FMCT, irreversibility and reduced operational status.
• The record since is one of repeated failure, and the pattern is diagnostic.
  • 2005 collapsed over procedure and retreat from the 2000 commitments; 2010 succeeded with a sixty-four-point action plan and a mandate for a Middle East zone conference never held.
  • 2015 failed on the Middle East zone; 2022 failed when Russia blocked consensus over the Zaporizhzhia plant in occupied Ukraine.
  • The Eleventh Review Conference, closing in New York in May 2026, again ended without a substantive outcome, deadlocked chiefly over whether Iran’s safeguards non-compliance would be named, with further disputes over North Korea and Ukraine’s nuclear facilities. It was the third consecutive failure, three months after the last strategic arms treaty expired.
• The review cycle is the only forum in which the non-nuclear majority can hold the five to account, and its serial breakdown removes the valve through which the disarmament grievance was vented.

The Verification and Supply-Side Architecture

IAEA safeguards and the Additional Protocol

• Safeguards are the regime’s verification instrument, and their evolution is a history of learning from failure.
  • INFCIRC/66-type safeguards are item-specific, covering a named facility or transferred item and nothing else — the model applying to states outside the NPT, including India.
  • INFCIRC/153-type safeguards are comprehensive, covering all nuclear material in all peaceful activities, and are obligatory for every non-nuclear-weapon state party.
  • The Additional Protocol of 1997 adds expanded declarations, access to any location on a site, environmental sampling and short-notice access.
• It exists because of a specific shock: after the 1991 Gulf War, inspectors found that Iraq had built a large clandestine enrichment programme alongside declared, fully safeguarded facilities.
  • Comprehensive safeguards had verified that declared material was where Iraq said, and had been blind to everything else. Material accountancy is not the same as detecting a hidden programme.
  • The distinction between correctness — the declaration is accurate — and completeness — it includes everything — became the Agency’s organising concept, and only the Additional Protocol supports a broader conclusion on completeness.
• The system’s weaknesses are structural: the IAEA has no enforcement power, depends on state cooperation and national intelligence, and can have its access withdrawn. Its record is mixed — it exposed Iraq’s programme, verified South Africa’s dismantlement and policed the Iran deal, but failed to detect North Korea’s diversion in time.

The test-ban chain

• Ending testing has been pursued since India’s 1954 proposal for a standstill on testing, and the instruments accumulated piecemeal.
  • The Partial Test Ban Treaty (1963), signed after the Cuban crisis, banned tests in the atmosphere, outer space and under water, leaving underground testing permitted; India was an original signatory. Its achievement was environmental, not strategic.
  • The Threshold Test Ban Treaty (1974) and Peaceful Nuclear Explosions Treaty (1976), both bilateral, capped underground yields at 150 kilotons and regulated explosions outside test sites.
• The Comprehensive Nuclear-Test-Ban Treaty, adopted by the General Assembly in 1996, prohibits any nuclear explosion, anywhere, by anyone.
  • Its rationale is that a total ban constrains both horizontal and vertical proliferation: entrants cannot validate a design, and established powers cannot confidently develop new types of warhead.
  • Its International Monitoring System — seismic, hydroacoustic, infrasound and radionuclide stations feeding an international data centre, with provision for on-site inspection — is the most elaborate verification system ever built, operated provisionally by the Preparatory Commission. It detected all six North Korean tests.
• The treaty has 187 signatories and 178 ratifications and is not in force, because Article XIV requires ratification by all 44 “Annex 2” states.
  • Eight have not ratified: China, Egypt, Iran, Israel and the United States signed but did not ratify; India, Pakistan and North Korea have not signed.
  • The American Senate rejected ratification in 1999 and no administration has resubmitted it; Russia de-ratified in November 2023, matching the American position while pledging not to test first.
  • In October 2025 the American President said he had instructed the Department of War to “start testing our Nuclear Weapons on an equal basis”, and the dispute over whether this means explosive or subcritical tests is unresolved.
    • No American explosive test has occurred since 1992, and the global moratorium — broken only by North Korea, most recently in 2017 — holds, but visibly contingently.
• India’s objection was never to a test ban as such.
  • The treaty is not genuinely comprehensive, because sub-critical experiments, hydrodynamic testing and simulation let advanced powers maintain and refine arsenals without explosions.
  • It would therefore have frozen the hierarchy rather than dismantled it, and India demanded a time-bound framework for global disarmament as the condition of signing.

“India will not sign this unequal treaty. Not now, not ever.” — Arundhati Ghose

The FMCT and the Conference on Disarmament

• The Fissile Material Cut-off Treaty would ban production of fissile material for weapons, acting as a cap that becomes a ceiling: warheads could be retained, but no new material made, so arsenals could only shrink.
• The Shannon Mandate of 1995 authorised negotiation of a “non-discriminatory, multilateral and internationally and effectively verifiable” treaty and left the question of existing stocks to the negotiations.
• Negotiations have never begun, because the Conference on Disarmament works by consensus.
  • Pakistan blocks the treaty, on the candid ground that a production cut-off would lock in India’s larger unsafeguarded stock.
  • The stocks question is the real dispute: the five have largely ceased production and hold large reserves, so a production-only ban costs them nothing.
  • The Conference has adopted no substantive programme of work for most of three decades — the clearest evidence that the multilateral disarmament machinery has stalled.
• India supports an FMCT that is universal, non-discriminatory and internationally verifiable, and committed to participating in negotiations under the 2008 arrangements, while continuing to produce material meanwhile.

Export-control regimes

Regime	Founded	Purpose	India
Zangger Committee	1971	Interprets NPT Article III(2) — the trigger list of items requiring safeguards on export	Not a member
Nuclear Suppliers Group	1975, in direct response to India’s 1974 test	Guidelines for nuclear and dual-use exports; decides by consensus; 48 governments	Not a member; holds the 2008 waiver; application blocked by China since 2016
Missile Technology Control Regime	1987	Restricts missiles able to deliver 500 kg to 300 km	Member, June 2016
Wassenaar Arrangement	1996, successor to COCOM	Conventional arms and dual-use goods and technologies	Member, December 2017
Australia Group	1985	Chemical and biological precursors and equipment	Member, January 2018
Hague Code of Conduct	2002	Pre-launch notification of ballistic missile and space-launch tests; some 145 subscribers	Subscriber, 2016
Proliferation Security Initiative	2003	An activity, not an organisation — coordinated interdiction of WMD shipments	Not a participant

• These are supplier cartels, not treaties — informal, non-binding, consensus-based and outside the UN.
  • Recipients read them as a second layer of discrimination imposed after they accepted the NPT bargain; suppliers reply that dual-use technology cannot be released unconditionally.
• The NSG is the pivotal case: it was created because of India and has now exempted India, which raises the question of what principle it applies.
  • China’s stated position is that the group must first agree a criteria-based approach for all non-NPT applicants, a two-step formula keeping Pakistan’s application procedurally alongside India’s.
  • India’s counter is that the NSG is an export-control mechanism, not a non-proliferation treaty, that France was a founding member in 1975 while outside the NPT, and that India has never transferred sensitive technology.

Nuclear-weapon-free zones and the spatial regime

Instrument	Year	Coverage
Antarctic Treaty	1959	Bans nuclear explosions and waste disposal in Antarctica
Outer Space Treaty	1967	Bans nuclear weapons in orbit or on celestial bodies
Treaty of Tlatelolco	1967	Latin America and the Caribbean — the first populated-region zone
Seabed Treaty	1971	Bans emplacement of nuclear weapons on the seabed
Treaty of Rarotonga	1985	South Pacific
Treaty of Bangkok	1995	Southeast Asia, extending to exclusive economic zones — why the nuclear powers have not signed its protocols
Treaty of Pelindaba	1996	Africa, following South Africa’s dismantlement
Treaty of Semipalatinsk	2006	Central Asia, binding states that once hosted Soviet weapons
Mongolia	2000	Single-state nuclear-weapon-free status, recognised by the General Assembly

• Zones do two things: bar members from acquiring, and bar the stationing, transit and testing of others’ weapons — the second always harder to secure, since the nuclear powers attach reservations to the protocols.
• Their political weight is that over a hundred states, covering almost the whole Southern Hemisphere, have accepted a legal status the nuclear powers reject for themselves.
• The conspicuous absence is the Middle East: on the agenda since 1974 and made the price of indefinite extension in 1995, the zone remains unrealised because of the Israel–Iran deadlock.
• The Korean Peninsula Denuclearisation Declaration of 1992, in which both Koreas renounced enrichment and reprocessing as well as weapons, was Asia’s most ambitious regional instrument and collapsed within a year.

Guarding against non-state actors

• Security Council Resolution 1540 (2004), adopted under Chapter VII after the Khan network was exposed, obliges all states to criminalise WMD acquisition by non-state actors and to enforce export controls, physical protection and accounting.
  • Its innovation is that the Council legislated general obligations rather than addressing a situation, which some states criticised as exceeding its competence.
• The Proliferation Security Initiative (2003) organises interdiction — boarding, inspection and seizure of suspect cargoes under national law and bilateral shipboarding agreements — and is limited by flag-state consent.
• The Convention on the Physical Protection of Nuclear Material and its 2005 Amendment, in force since 2016, extend protection obligations from material in international transport to material in domestic use and storage and to facilities themselves.
• The Nuclear Security Summits (2010–2016) produced measurable results — elimination and repatriation of highly enriched uranium, reactor conversions, ratification of the CPPNM amendment — but addressed material security, not weapons, and were discontinued after 2016 without a successor of comparable weight.
• The International Convention for the Suppression of Acts of Nuclear Terrorism (2005) criminalises possession and use of radioactive material with intent to cause harm.

Missile Proliferation and the Diffusion of Delivery Technology

Warheads without delivery vehicles are a laboratory achievement. The strategic consequence of nuclear diffusion depends on the parallel and faster diffusion of the means to deliver them — governed by far weaker instruments, because there is no missile equivalent of the NPT.

Why delivery is the decisive variable

• Range converts a local capability into a global concern: a state with a fission device and short-range aircraft threatens neighbours; with an intercontinental missile it alters calculations on other continents.
• Delivery determines doctrine. Fixed, liquid-fuelled missiles are vulnerable and invite early use; mobile, solid-fuelled, canisterised systems permit a relaxed posture and a genuine second strike.
• Space launchers and ICBMs are the same technology differently configured, so a civil space programme is a lawful route to delivery capability and cannot be prohibited.
• Warning time collapses between neighbours, removing deliberation and forcing pre-delegation and launch-on-warning — the technical root of Sagan’s accident argument.
• Missiles are first-strike-friendly, since there is no effective defence against a saturating attack, which is why missile defence and missile proliferation form one destabilising loop.

Type	Characteristics	Significance
Ballistic missiles	Boosted then unpowered on a predictable arc; short to intercontinental; increasingly solid-fuelled, mobile and MIRV-capable	The classic strategic system; detectable in flight, which makes warning and arms control technically possible
Cruise missiles	Powered throughout, air-breathing, terrain-following; launched from air, sea, submarine or land	Cheap, accurate, hard to detect, and dual-capable — creating warhead ambiguity in a crisis
Hypersonic systems	Glide vehicles and cruise missiles manoeuvring above Mach 5 in the atmosphere	Compress decision time, defeat existing defences, blur conventional and nuclear strike
Armed drones	Cheap, attritable, exportable, increasingly autonomous	Not nuclear delivery, but the fastest-diffusing strike technology in existence, reaching state and non-state actors alike

The MTCR and its limits

• The Missile Technology Control Regime, founded in 1987 by seven states and now with 35 partners, is the principal instrument.
  • Category I — complete systems delivering a payload of 500 kg to 300 km, and their production facilities — carries a strong presumption of denial; Category II covers dual-use components, propulsion, guidance and materials case by case.
  • The threshold was chosen because it approximates a first-generation warhead on a militarily useful missile.
• Its weaknesses are structural.
  • It is a political understanding, not a treaty — no legal obligation, no verification, no penalty for transfer.
  • It is a supplier cartel with critical non-members: China is outside and repeatedly accused of transfers, while North Korea, Iran and Pakistan are outside and trade with each other.
  • It does not address indigenous development, and the states of greatest concern now build rather than buy.
  • The threshold is obsolescent: miniaturised warheads fall below 500 kg, and cruise missiles, drones and space launchers fit awkwardly in a ballistic-missile framework.
  • Application has been politically inconsistent where alliance interests dictated.
• The Hague Code of Conduct (2002) adds a norm and transparency layer, committing some 145 subscribers to pre-launch notifications and annual declarations.
  • Its limits mirror the MTCR’s: politically binding only, covering ballistic missiles and space launchers but not cruise missiles, with China, Iran, North Korea, Pakistan and Israel outside.
• India joined the MTCR in 2016 and subscribes to the Code — the combination that gave it standing in the export-control architecture while remaining outside the NPT, and a say over China’s admission.

The drone extension

• Armed unmanned systems have diffused faster than any strategic technology in decades because they are cheap, exportable and covered by no dedicated regime.
  • Turkish, Chinese and Iranian systems have gone to states no supplier cartel would have licensed for manned strike aircraft, and non-state actors have used commercial quadcopters and improvised loitering munitions in Iraq, Syria, Yemen and Ukraine.
  • The 2020 American reinterpretation treating slower armed drones as Category II rather than Category I loosened the strongest control that existed.
• The convergence risk connects back to the nuclear question: autonomy, artificial intelligence and cheap precision strike are entering the same operational space as nuclear command and control, and no instrument governs the intersection.

The Bilateral Arms-Control Chain and Its Collapse

Instrument	Year	Effect	Fate
Hotline Agreement	1963	Direct communications link after the Cuban crisis	Operative
ABM Treaty	1972	Limited strategic missile defences to preserve mutual vulnerability — the treaty codifying MAD	US withdrawal, 2002
SALT I Interim Agreement	1972	Froze numbers of strategic ballistic missile launchers	Expired 1977
SALT II	1979	Ceilings on delivery vehicles and MIRVed systems	Signed, never ratified after Afghanistan
INF Treaty	1987	Eliminated an entire class — all ground-launched missiles of 500–5,500 km	Both parties withdrew, 2019
START I	1991	Verified cuts to 6,000 accountable warheads, with on-site inspection	Expired 2009
START II	1993	Would have banned MIRVed ICBMs	Never in force
SORT	2002	Cut deployed warheads to 1,700–2,200 — three pages, no verification	Superseded 2011
New START	2010	1,550 deployed strategic warheads, 700 delivery vehicles, with inspections and data exchanges	Extended 2021; Russian participation suspended 2023; expired 5 February 2026
Open Skies	1992	Unarmed observation overflights	US withdrew 2020; Russia 2021

• The chain’s logic was cumulative: each agreement built verification practice and habits of dialogue that made the next possible. Its crown jewel was intrusive verification, which made both sides’ forces legible to each other — a greater contribution to stability than the numerical limits.
• The unravelling has been sequential and is now essentially complete.
  • The chain’s most productive phase depended on one leader: Gorbachev‘s new thinking converted arms control from ceilings into elimination, delivering the INF Treaty and the framework for START, which is a reminder that the machinery has always been driven by political relationships rather than the reverse.
  • 2002: the United States left the ABM Treaty; Russia read it as a bid to negate its deterrent and began the exotic systems it later unveiled.
  • 2019: both parties left the INF Treaty, over an alleged Russian ground-launched cruise missile and allegedly treaty-capable American launchers in Europe.
  • 2020–21: withdrawal from Open Skies removed a transparency instrument of particular value to smaller European states.
  • 2023: Russia suspended participation in New START after the invasion of Ukraine, ending inspections while declaring it would observe the numerical limits.
  • September 2025: Russia offered to observe New START’s central limits for a year beyond expiry if reciprocated; no agreement followed.
  • 5 February 2026: New START expired. For the first time since 1972, no negotiated limit or verification regime governs American and Russian strategic arsenals.
    • Moscow has since said it will hold to the former limits only while Washington does; Washington insists a successor be a new and modernised agreement, preferably including China, which Beijing declines on the ground that its arsenal is a fraction of the other two.
• The consequences are not primarily about numbers.
  • Verification has ended, so each side estimates the other from national means alone — historically a source of worst-case assumption and over-building.
  • The trilateral problem has no precedent: Cold War arms control was bilateral and symmetrical, and a lagging third party has weak incentive to accept limits.
  • Uploading capacity is large: both hold substantial non-deployed warheads returnable to delivery vehicles quickly, so a race needs no new production.
  • The expiry came three months before the NPT Review Conference, removing any remaining case that Article VI was being performed.

Proliferation in Practice: The Cases

North Korea: the only state to leave the treaty and build

• North Korea is the regime’s single unambiguous failure, and each stage was a choice its opponents failed to prevent.
  • It joined the NPT in 1985 for Soviet reactor assistance, delayed safeguards, and by 1992 inspections found discrepancies indicating undeclared plutonium separation.
  • It announced withdrawal in 1993, suspended the notice a day before it took effect, and used the crisis to extract the Agreed Framework (1994) — a freeze of Yongbyon for light-water reactors, heavy fuel oil and steps toward normalisation.
  • The Framework collapsed in 2002 over a covert uranium enrichment programme the plutonium-focused deal had not covered, and it withdrew from the NPT in January 2003, the only state ever to do so.
  • The Six-Party Talks produced the September 2005 Joint Statement, the most complete commitment ever obtained to abandon all weapons and programmes, which unravelled over verification and financial sanctions.
  • It tested in 2006, 2009, 2013, twice in 2016 and in September 2017, the last with a claimed thermonuclear yield of the order of 100 kilotons, and has since demonstrated intercontinental-range, solid-fuelled and submarine-launched systems.
• The summit diplomacy of 2018–19 was the most direct engagement attempted, and its failure is instructive.
  • The Singapore Summit (June 2018) produced a commitment to work toward “complete denuclearization of the Korean Peninsula” — a phrase never defined, read in Washington as North Korean disarmament and in Pyongyang as including the American umbrella over the South.
  • It contained no inventory, timeline, verification mechanism or sequencing, which is why it produced atmospherics and suspended exercises but no dismantlement.
  • The Hanoi Summit (February 2019) collapsed over sequencing: Yongbyon offered for substantial sanctions relief, the undeclared enrichment sites and the weapons demanded in return; the Panmunjom meeting (June 2019) briefly restored contact before working-level talks broke down, and diplomacy has been dormant since.
• The present position is consolidation, not negotiation.
  • Estimates put the arsenal at roughly fifty to sixty warheads, with fissile material for more, and Pyongyang publicised a uranium enrichment facility in 2024 as a deliberate disclosure.
  • It legislated a nuclear-use doctrine in 2022 permitting pre-emptive use, and wrote its nuclear-weapons-state status into its constitution in 2023 to make the commitment irreversible.
  • It has abandoned unification, redefined the South as a hostile state, and deepened alignment with Russia — which cost the sanctions regime its expert monitoring panel to a veto.
• The lessons are uncomfortable: Article X is a real exit with no consequence attached; sanctions did not prevent acquisition by a state willing to absorb extraordinary costs; and no state that already had weapons has been negotiated out of them.

Iran: the deal, its destruction and the aftermath

• Iran is the case where the regime’s instruments worked, were abandoned, and cannot now be reassembled. Undeclared facilities at Natanz and later Fordow were exposed in 2002 and 2009, and the IAEA found non-compliance, referring the file to the Security Council in 2006.
• The Joint Comprehensive Plan of Action (2015) was the most technically detailed non-proliferation agreement ever concluded, and its terms are the argument.
  • Enrichment capped at 3.67% U-235 for fifteen years, with the low-enriched stockpile capped at 300 kilograms and centrifuges cut from around 19,000 to 5,060 first-generation machines; Fordow converted to research with no enrichment.
  • The Arak heavy-water reactor redesigned so it could not produce weapons-grade plutonium, with no reprocessing for fifteen years.
  • Application of the Additional Protocol plus continuous monitoring of centrifuge manufacturing and uranium mining — the most intrusive verification any state has accepted.
  • The design objective was a breakout time of at least twelve months, against a pre-deal estimate of two to three months.
  • Sunset clauses phased out the principal restrictions between 2025 and 2031 — read by supporters as buying a decade plus a permanent Additional Protocol, by critics as a legal path to an industrial programme.
  • Snapback was the enforcement innovation: Resolution 2231 let any participant trigger automatic reimposition of pre-2015 UN sanctions without risk of veto, since a resolution to continue relief would itself be vetoable.
• The American withdrawal in May 2018 removed Iran’s incentive to comply, and the unravelling followed a predictable sequence.
  • Enrichment rose to 20% from 2019 and to 60% U-235 from 2021, a level with no plausible civil use.
  • IAEA access was reduced, cameras removed and experienced inspectors de-designated, breaking the Agency’s continuity of knowledge.
  • In June 2025 the Board found Iran in non-compliance with its safeguards obligations for the first time in two decades.
• The collapse of 2025 is the decisive recent development.
  • In June 2025 Israel and then the United States struck Natanz, Fordow and Isfahan — the first direct attack on the safeguarded facilities of an NPT party.
  • Iran suspended cooperation with the IAEA in July 2025; inspectors withdrew and access to most declared sites ended.
  • On 28 August 2025 the three European parties triggered snapback, and on 28 September 2025 the full pre-2015 UN sanctions regime was reimposed; the JCPOA’s UN framework lapsed with Resolution 2231 in October 2025.
  • As of 2026 the Agency cannot verify the size, composition or whereabouts of Iran’s enriched-uranium stockpile or confirm whether enrichment has resumed; most of the 60% material is believed to remain in the tunnel complex at Isfahan.
• The analytical points matter more than the chronology.
  • Verification is the deal’s real asset. What the withdrawal destroyed was not centrifuges but knowledge, and knowledge is harder to rebuild than hardware.
  • Military action buys time and destroys transparency, removing the inspectors who could have said by how much, and strengthening the domestic argument for weaponisation.
  • Enforcement now runs outside the treaty: snapback and airstrikes are both responses to a treaty violation, and neither is a treaty mechanism.
  • The precedent cuts across the regime: a state that accepted the most intrusive safeguards in history was attacked; a state that left the treaty and built weapons was not.

Israel: opacity as policy

• Israel is universally assessed to hold nuclear weapons — around ninety warheads — and has never confirmed or denied it.
  • The doctrine of opacity, or amimut, is expressed as a promise not to be the first to introduce nuclear weapons into the Middle East, read as meaning not the first to test or declare.
  • Its purpose is existential deterrence without the political consequences of declaration: no arms race triggered by an announcement, no forced sanctions, no formal collapse of the regional position.
  • The Dimona reactor was built with French assistance from the late 1950s, was never placed under comprehensive safeguards, and Israel has never joined the NPT.
• The Vanunu affair is how the scale became known: Mordechai Vanunu, a Dimona technician, gave photographs and a description of the plutonium-separation facility to a British newspaper in 1986, and was abducted from Rome and imprisoned for eighteen years.
• The case is the clearest instance of selective enforcement, and Israeli capability is the standing justification for Arab and Iranian refusal to treat the Middle East zone as someone else’s business.
  • It also established counter-proliferation by force: the Osirak strike of 1981, the 2007 strike on Syria’s Al-Kibar and the 2025 strikes on Iran are one continuous policy, the Begin Doctrine.

Pakistan and the A.Q. Khan network

• Pakistan’s programme is the clearest example of the security model: begun after the defeat of 1971, accelerated after India’s 1974 test, pursued through the enrichment route.
  • A.Q. Khan brought centrifuge designs and supplier lists from a European enrichment consortium, and Pakistan built on highly enriched uranium rather than plutonium, testing within a fortnight of Pokhran-II in May 1998.
• Its doctrine is the deliberate mirror of India’s.
  • No no-first-use: Pakistan reserves first use because the weapons offset conventional inferiority, not merely nuclear attack.
  • Full-spectrum deterrence extends the threat across strategic, operational and battlefield tactical ranges, with the short-range Nasr system designed to make an Indian conventional advance nuclear-risky at once.
  • The consequence is a lowered threshold, delegation pressure on custody in a crisis, and precisely the organisational risks Sagan identifies.
• The A.Q. Khan network was the most consequential proliferation event since the Manhattan Project, because it was private, transnational and market-based.
  • It supplied centrifuge designs, components and in at least one case a weapon design to Libya, Iran and North Korea, using manufacturers across Europe and Asia and transhipment through Dubai and Malaysia.
  • It was exposed by the interdiction of the ship BBC China carrying centrifuge components to Libya in 2003, which precipitated Libya’s disarmament.
  • It proved that supply-side controls built around states could be defeated by a network, which is exactly why Resolution 1540 imposes obligations on states to control non-state actors in their jurisdictions.

South Africa, Libya, Iraq, Syria and Ukraine

• South Africa is the only state to build and then dismantle: six gun-type devices and part of a seventh, destroyed between 1989 and 1991, followed by accession to the NPT and comprehensive safeguards — the only completeness verification the IAEA has ever performed.
  • The strategic rationale disappeared with Cuban withdrawal from Angola and the end of the Cold War; the transition to majority rule meant the outgoing government preferred destruction to handover; and disarmament purchased reintegration, sanctions relief and a leadership role in the Treaty of Pelindaba. It confirms Solingen’s model precisely.
• Libya (2003) abandoned a Khan-supplied programme for sanctions relief and normalisation; the overthrow and killing of Gaddafi in 2011 converted a non-proliferation success into the most powerful argument against disarmament available to any state, cited explicitly by Pyongyang.
• Iraq reshaped verification twice: the 1991 discovery of a clandestine programme under comprehensive safeguards produced the Additional Protocol, and the 2003 invasion premised on weapons that did not exist destroyed public confidence in intelligence-led counter-proliferation.
• Syria’s suspected reactor at Al-Kibar, built with North Korean assistance and destroyed in 2007, showed both that undeclared construction remained possible and that the response was again force rather than institutional action.
• Ukraine carries the sharpest consequence for the bargain’s future.
  • On the Soviet collapse Ukraine held the third-largest arsenal in the world on its territory, though not operational control; Belarus and Kazakhstan were comparable.
  • Under the Budapest Memorandum (1994) all three transferred the weapons to Russia and acceded to the NPT as non-nuclear-weapon states, in exchange for assurances from Russia, the United States and the United Kingdom to respect their sovereignty and existing borders and to refrain from force.
  • The annexation of Crimea in 2014 and the invasion of 2022 were carried out by a guarantor, and the others did not reverse them.
  • The lesson drawn worldwide is the one the regime can least afford: the state that gave up nuclear weapons for security assurances was invaded, and the states that kept them were not.

The cascade question

• The proliferation cascade — the fear since the 1960s that each new nuclear state triggers several more — has repeatedly failed to materialise, and is now more plausible than at any time since.
  • South Korea is the most advanced case: public majorities have supported an indigenous deterrent for years, and support now spans the political spectrum rather than the right alone.
    • The binding constraint has been the 123 Agreement denying enrichment and reprocessing consent; negotiations over enrichment for naval propulsion, and an American signal of openness in late 2025, removed part of it.
  • Japan holds the largest separated plutonium stock of any non-nuclear-weapon state, plus enrichment, reprocessing and a launcher industry. Its latency is deliberate, and its calculation turns entirely on the credibility of the American guarantee.
  • Saudi Arabia has said it will match any Iranian capability, and negotiations over an American civil nuclear agreement have turned on whether Riyadh must accept the “gold standard” forswearing of enrichment and reprocessing that the United Arab Emirates accepted.
  • Poland has sought participation in NATO nuclear sharing; Germany has seen debate about a European deterrent unthinkable a decade ago; Turkey’s leadership has publicly questioned why others may hold weapons and Turkey may not.
• What has changed is not capability, which was always present, but three simultaneous shifts: doubt about American extended deterrence, the demonstrated fate of Ukraine, and the visible collapse of arms control.
  • The countervailing forces — alliance benefits, economic integration, domestic opposition and stigma — remain strong, which is why the cascade is a risk and not a forecast.
  • But the regime’s deterrent against acquisition has always been the certainty of cost, and the states now weighing the question are precisely those the United States would find hardest to punish.

The Disarmament Argument

Article VI and the ICJ

• Article VI is the legal hinge, and its exact wording controls the whole argument.

“Each of the Parties… undertakes to pursue negotiations in good faith on effective measures relating to cessation of the nuclear arms race at an early date and to nuclear disarmament.” — Article VI, Nuclear Non-Proliferation Treaty

• The obligation is therefore to negotiate, not to disarm, and it extends further — to “a treaty on general and complete disarmament under strict and effective international control”, a phrase the weapon states have never treated as operative.

• The nuclear-weapon states read it as an obligation of conduct, not of result; the majority reads it, with the 2000 “unequivocal undertaking”, as a commitment to elimination whose performance is measurable and absent.

The humanitarian initiative and the TPNW

• The prohibition treaty was produced by a deliberate reframing: moving the argument from strategic stability, where the nuclear states are the experts, to humanitarian consequences, where every state and civil society has standing.
  • Conferences at Oslo (2013), Nayarit (2014) and Vienna (2014) assembled the medical, climatic and famine evidence and issued the Humanitarian Pledge, endorsed by well over a hundred states.
  • The International Campaign to Abolish Nuclear Weapons coordinated the civil-society side and won the Nobel Peace Prize in 2017 — the epistemic-community mechanism working as Haas describes.
• The Treaty on the Prohibition of Nuclear Weapons was adopted in July 2017 by 122 states and entered into force on 22 January 2021 after the fiftieth ratification.
  • It prohibits developing, testing, producing, acquiring, possessing, stockpiling, transferring, stationing, using or threatening to use nuclear weapons, and assisting any of it.
  • It adds obligations the earlier instruments lack: victim assistance and environmental remediation for those affected by use and testing.
  • It offers a possessor two accession routes — destroy then join, or join then destroy under a legally binding time-bound plan.
  • It now has around 75 states parties and about a hundred signatories, with Meetings of States Parties in Vienna (2022) and New York (2023 and March 2025), and its First Review Conference set for New York from 30 November to 4 December 2026.
• No nuclear-armed state and no NATO member has joined, which is the fact defining every argument about it.

Question	TPNW supporters	The step-by-step school
Legal effect	A categorical prohibition completing the WMD architecture beside the chemical and biological conventions	A treaty binding only those who never had the weapons changes nothing about arsenals
Relation to the NPT	Implements Article VI — it is the “effective measure” the NPT requires	Undermines the NPT by creating competing standards and a rival forum
Mechanism	Stigmatisation — delegitimise the weapons and financing, doctrine and politics follow, as with landmines	Disarmament needs the security conditions that make it safe; a security requirement cannot be legislated away
Verification	Provides for safeguards and a competent authority designated on accession	Thin against the real challenge of dismantling arsenals
Evidence	Divestment by pension funds; parliamentary support growing inside umbrella states	Zero warheads eliminated since entry into force; arsenals growing

Global Zero, the Four Horsemen and the modernisation reality

• The disarmament case is not confined to the non-nuclear world. George Shultz, William Perry, Henry Kissinger and Sam Nunn argued from 2007 that deterrence was becoming less effective and more hazardous as capabilities spread, and endorsed elimination together with a ladder of practical steps.
  • Their argument is prudential rather than moral: nuclear weapons are a declining asset in a world of many possessors, terrorist demand and short warning times.
  • Global Zero and the 2009 Prague speech carried the goal into policy, and New START was its principal product.
• The counter-fact is that every one of the nine possessors is modernising.
  • The United States is recapitalising the whole triad — new missile, bomber, submarine and warhead programmes — at a cost estimated in the trillions over three decades; Russia has fielded new missiles and exotic systems framed as answers to missile defence.
  • China is expanding fastest, at around 620 warheads at the start of 2026, with new silo fields and a developing triad; the United Kingdom has raised its declared ceiling, France is modernising both legs, and India, Pakistan, Israel and North Korea are all growing or improving.
  • Global inventories stand at roughly 12,200 warheads, of which about 9,700 are in military stockpiles and some 4,000 deployed — far below the Cold War peak above 70,000, but with the direction of travel now reversed for the first time since the 1980s.

The number of warheads fell for thirty years because political relationships improved, not because treaties compelled it — and the same causation is now running the other way.

New Technologies and the Next Nuclear Age

• Missile defence remains the destabiliser at the centre of the system, for the reason the ABM Treaty recognised: a defence that might work degrades confidence in the second strike, so the rational answer is more warheads, more decoys, more MIRVs. Defence produces offence, and the effect is now trilateral.
• Hypersonic glide vehicles and cruise missiles compress decision time and obscure a launch’s target and payload, undermining the warning architecture on which launch decisions rest.
• Counterspace capabilities threaten the satellites providing early warning and nuclear command and control; an attack on warning satellites in a crisis is indistinguishable from preparation for a first strike, which makes anti-satellite weapons escalatory even when used conventionally. Tests by the United States, Russia, China and India have all added to the problem.
• Cyber threats to nuclear command and control are the least governed risk: spoofed early warning, corrupted decision-support data, and loss of confidence in one’s own systems in a crisis. Entanglement compounds it, since the same sensors and networks serve conventional and nuclear missions.
• Artificial intelligence in nuclear decision-making poses the sharpest new question.
  • Applied to warning fusion and target recognition it can reduce human error, while introducing automation bias, opacity and compression of the deliberation time on which every historical near-miss depended.
  • Those near-misses were resolved by individual officers overriding the machine, which is the exact judgement an automated pipeline removes.
  • A narrow emerging consensus that human control must be retained over nuclear-use decisions has been asserted bilaterally and in declarations, with no verification and no treaty.
• Nuclear terrorism remains the low-probability, unbounded-consequence residual.
  • The pathways are a stolen weapon, an improvised device from stolen material, a radiological dispersal device, or sabotage of a facility — the last turned from scenario into operational risk by the shelling around Zaporizhzhia.
  • No group has obtained a weapon, which is the best evidence the post-2001 security regime worked; its weakness is that the political process sustaining it lapsed after 2016.

India and the Global Non-Proliferation Order

India’s relationship with the nuclear order is the most instructive single case in the subject, because India has occupied every available position: the earliest advocate of universal disarmament, the first state outside the recognised five to test, the most articulate critic of the regime’s discrimination, and finally the only state granted an exemption from rules it refused to sign.

Origins, Nehru’s advocacy and the option kept open

• India’s nuclear establishment predates almost every other, and was built as a development project with the weapons option deliberately left open.
  • The Atomic Energy Commission was established in 1948 under the Atomic Energy Act, within a year of independence and before most industrial states had civil programmes.
  • Homi Bhabha designed the three-stage programme around India’s resources: pressurised heavy-water reactors on natural uranium, fast breeder reactors on the plutonium they produce, and a third stage exploiting India’s very large thorium reserves. Its object was energy self-reliance, and its by-product was mastery of the full fuel cycle.
  • Nehru treated the bomb as morally unacceptable and authorised no weapons programme, while declining every proposal that would have foreclosed the option — including the Baruch Plan, which India opposed as prohibiting national research and development.
  • India was the first state to propose, in 1954, a standstill agreement on nuclear testing, a proposal that fed into the Partial Test Ban Treaty nine years later.
  • The 1962 war with China and the Chinese test of October 1964 removed the strategic basis of abstention; India sought guarantees from both superpowers and obtained nothing usable.

“Atomic weapons give a State possessing them in adequate numbers a deterrent power against attack from a much stronger State.” — Homi Bhabha

• Shastri authorised only theoretical work on a subterranean explosion for peaceful purposes in 1964; the deaths of both Shastri and Bhabha in 1966 and the accession of Indira Gandhi shifted the balance.

1974: the peaceful nuclear explosion and the creation of the NSG

• India refused to sign the NPT in 1968, on grounds it still gives: the treaty was discriminatory, imposed obligations only on the have-nots, contained no disarmament timetable, and legitimised five arsenals by reference to a calendar date.
• On 18 May 1974 India tested a fission device of roughly eight kilotons at Pokhran, describing it as a peaceful nuclear explosion and known informally as Smiling Buddha, from the coded message by which Raja Ramanna, the project head, reported success.
  • The description was legally available: the NPT itself referred to the benefits of peaceful nuclear explosions, and both superpowers had conducted them.
  • Politically it demonstrated capability without declaring a weapons programme, preserving ambiguity and the disarmament position at once.
• The response was structural, which is why the test matters to the global regime and not only to India.
  • Canada terminated cooperation, and the United States treated the test as a breach of Atoms for Peace understandings.
  • The Nuclear Suppliers Group was created in direct response, extending controls beyond NPT parties to any recipient and eventually to dual-use items — the birth of the supply-side architecture as a permanent feature.
  • The American Nuclear Non-Proliferation Act of 1978 made full-scope safeguards a condition of supply, cutting off fuel for the Tarapur reactors the United States had itself built.
• India then did not test again for twenty-four years — it had crossed the threshold and declined to weaponise, a posture known as keeping the option open or recessed deterrence.
  • Rajiv Gandhi authorised weaponisation in the late 1980s in response to intelligence about Pakistan’s programme, while simultaneously tabling the most ambitious disarmament proposal ever put to the UN.

The Rajiv Gandhi Action Plan

• The Action Plan for a Nuclear-Weapon-Free and Non-Violent World Order, presented to the UN General Assembly’s Third Special Session on Disarmament in 1988, remains the only comprehensive disarmament plan tabled by a serving head of government.
  • It proposed a binding commitment to eliminate nuclear weapons in phases by 2010, within a time-bound framework rather than an open-ended obligation.
  • It set out a three-stage sequence — halting the arms race and freezing production, then substantial reductions with non-nuclear states accepting binding commitments not to acquire, then elimination — with verification at each stage.
  • It linked disarmament to a wider non-violent world order, including restructuring the international economic system and strengthening UN dispute settlement.
  • Its significance is the consistency it establishes: India objected not to non-proliferation but to non-proliferation unaccompanied by disarmament, and offered a concrete alternative rather than a refusal. Its rejection is generally treated as the moment restraint lost to capability.

The CTBT, the NPT extension and 1998

• The early 1990s combined the indefinite extension of the NPT in 1995 with a test ban that would have foreclosed India’s option permanently, and India read the combination as the closure of the order against it.
  • India participated in the CTBT negotiations and then blocked consensus in the Conference on Disarmament in 1996, objecting that the treaty banned explosions while permitting the sub-critical and simulation techniques by which advanced powers maintain arsenals, and contained no disarmament commitment.
  • India also objected to the Article XIV entry-into-force clause, which named India among the forty-four required ratifiers — a provision it called coercive.
• Pokhran-II, on 11 and 13 May 1998 as Operation Shakti, comprised five devices including a fission device, a low-yield device and a thermonuclear device, after which Vajpayee declared India a nuclear weapon state.
  • The strategic purpose was weaponisable design validation — placing warheads on the Prithvi and Agni missiles then entering service.
  • Pakistan tested within a fortnight; the Security Council responded with the unanimous Resolution 1172, calling on both to accede to the NPT as non-nuclear-weapon states, which neither has done or will do. Sanctions followed and were largely lifted within two years.
• The tests moved India from ambiguity to declared status, and with it from principled dissent to the need for a doctrine, a command system and a strategy for readmission.

The doctrine: 1999 draft and 2003 posture

• The Draft Report of the National Security Advisory Board (August 1999) set the framework; the Cabinet Committee on Security statement of January 2003 is the operative posture. Together they express a political rather than warfighting conception of nuclear weapons.
  • Credible minimum deterrence: an arsenal sufficient to inflict unacceptable damage and no larger, where “minimum” is a function of the strategic environment and “credible” imports survivability and demonstrated will.
  • No first use: weapons will be used only in retaliation against a nuclear attack on Indian territory or on Indian forces anywhere.
  • The chemical and biological caveat: India retains the option of nuclear retaliation against a major attack with biological or chemical weapons — the single qualification to NFU.
  • Massive retaliation: retaliation will be massive and designed to inflict unacceptable damage, deliberately foreclosing graduated response and denying an adversary any belief in limited nuclear use.
  • Non-use against non-nuclear-weapon states, an unconditional negative security assurance.
  • Civilian control through the Nuclear Command Authority, whose Political Council, chaired by the Prime Minister, alone can authorise use, with an Executive Council chaired by the National Security Adviser providing inputs and executing directives; custody is separated from the armed forces in peacetime.
  • Strict export controls, participation in FMCT negotiations, a continuing testing moratorium, and continued commitment to global, verifiable and non-discriminatory disarmament.
• The reading that best explains the doctrine’s shape treats it as an expression of political culture rather than strategic necessity.
  • Rakesh Sood describes India as a reluctant nuclear-armed state, and Raja Menon holds that the doctrine reflects India’s culture more than its strategic choices — an argument rooted in the non-violent character of the freedom struggle and in a tradition of restraint coupled with a capacity to coerce.
• The doctrine’s coherence rests on assured second strike, and the force has been built for it: a triad completed with INS Arihant and successors, canisterised and mobile Agni systems, and progressively longer-range and MIRV-capable missiles.
  • SIPRI’s January 2026 estimate places India’s inventory at around 190 warheads, with a small number assessed as deployed for the first time — a change from the historical posture of fully de-mated storage.

“The label on a nuclear weapon used for attacking India, strategic or tactical, is irrelevant from the Indian perspective.” — Shyam Saran

The no-first-use debate

• The case for retention is strategic rather than sentimental.
  • It defines the threshold clearly, which is what makes deterrence communicable; removing it injects ambiguity into a posture whose credibility depends on clarity.
  • It permits de-mated, low-readiness storage, reducing risks of accident, unauthorised use and pre-emption, and avoids the cost of alert infrastructure.
  • It places the onus of escalation on the adversary and removes India’s incentive to strike first, which in turn removes the adversary’s incentive to strike first for fear that India might.
  • K. Subrahmanyam‘s formulation is the most influential: deterrence is a matter of perception rather than posture, so a first-use declaration would be provocative without adding deterrent value, and might force an adversary to attack in order to avoid being attacked.
  • Shivshankar Menon cites China’s maintenance of no first use despite far greater asymmetry with the United States: what produces deterrence is survivability, not declaratory aggressiveness.
  • Rajesh Rajagopalan argues the logic is counter-intuitive but sound — a state facing no existential threat, and the strongest in its neighbourhood, has little use for first use, and India’s arsenal is anyway too modest for a disarming counterforce strike against two nuclear-armed neighbours.
  • Manpreet Sethi adds the economies: no expensive high-alert infrastructure, no forces kept mated and on trigger, a smaller burden of guarding against unauthorised use.
• The case against is made by serious figures.
  • Bharat Karnad argues NFU presupposes confidence in force survivability and competent crisis management, and that India’s institutional machinery supplies neither.
  • B.S. Nagal, a former commander-in-chief of the Strategic Forces Command, argues it is morally indefensible to accept large-scale destruction of one’s own population before responding.
  • P.K. Chari treated doctrines as work in progress rather than fixed, noting that a posture calibrated only against nuclear attack leaves India deterred at the conventional level while Pakistan’s posture deters India across the whole spectrum.
  • The signalling argument is that reserving first use would communicate resolve to a conventionally superior China cheaply. Harsh V. Pant notes the tension it exposes: India has reasons to want first-strike capability, and the actual state of its arsenal suggests it will not acquire them.

“No first use is relevant only for the country that has extreme confidence in the survivability of its nuclear forces.” — Bharat Karnad

• Official statements have kept the doctrine formally intact while eroding its edges.
  • A 2010 formulation of “no first use against non-nuclear weapon states”, read as implying that the assurance to nuclear-armed states was conditional; a 2014 manifesto promise to revise and update the doctrine; and a Defence Minister’s remark in 2016 that India should not bind itself.
  • In August 2019 the Defence Minister stated that India remained firmly committed to no first use but that what happens in future depends on the circumstances.
  • The Ministry of External Affairs has consistently confirmed no change, and the settled position is that the doctrine stands while periodic review is treated as legitimate.
• Massive retaliation is separately contested as not credible against a Pakistani tactical strike on Indian forces on Pakistani soil, since city-destroying retaliation invites the destruction of Indian cities. The official answer is that India does not recognise the tactical–strategic distinction at all.

Readmission: the civil nuclear deal, the waiver and the regimes

• After 1998 India held nuclear weapons, a real energy problem and no access to the fuel market. Domestic uranium was adequate for weapons and wholly inadequate for a large civil sector, so NSG restrictions bit on electricity rather than on the arsenal.
• The India–United States Civil Nuclear Agreement, launched by the joint statement of 18 July 2005, built a bespoke exception in four sequenced parts.
  • A separation plan: India designated facilities as civilian or military, safeguarding the former while the military programme remained wholly outside.
  • Amendment of American law: the Hyde Act of 2006 exempted cooperation with India from full-scope-safeguards requirements, permitting the bilateral 123 Agreement of 2007, named for the section of the American Atomic Energy Act governing cooperation agreements.
  • An India-specific IAEA safeguards agreement, approved in August 2008, of the item-specific type, with an Additional Protocol limited to safeguarded facilities.
  • An NSG waiver in September 2008, adopted by consensus, permitting trade with India despite its non-membership of the NPT and its refusal of full-scope safeguards.
• The result made India the only state with nuclear weapons outside the NPT permitted to conduct civil nuclear commerce — the largest exception the regime has ever made.
  • Supporters read it as the regime absorbing an unavoidable reality and converting an outsider into a stakeholder: safeguards on civilian facilities, a reaffirmed testing moratorium, harmonised control lists, a commitment to FMCT negotiations.
  • Critics argue it rewarded a state that tested, weakening the incentive structure the treaty depends on, and that its logic could not be confined — Pakistan demanded parity, and China responded with reactor sales to Pakistan on a contested grandfathering claim.
  • The Indian assessment is that the deal was geopolitical rather than commercial.
    • C. Raja Mohan reads it as recognition of India as a major power that legitimised its weapons; Rajeswari Rajagopalan attributes it to acknowledgement of India’s non-proliferation record; Sumit Ganguly locates the American motive in balancing a rising China; and Jaishankar has called it the turning point that raised India’s global standing.
• Membership of MTCR (2016), Wassenaar (2017) and the Australia Group (2018), with subscription to the Hague Code of Conduct, consolidated the position without resolving it.
  • The NSG remains blocked by China’s insistence on a criteria-based approach for all non-NPT applicants before any individual admission, which keeps Pakistan’s parallel application alive. The practical cost of exclusion is modest since the waiver already delivers the trade; the cost is status, which is what the argument has always been about.
• India’s position on the remaining instruments is consistent throughout.
  • NPT: will not accede as a non-nuclear-weapon state, and no other category exists; India seeks recognition as a responsible state with advanced nuclear technology, a formulation that avoids the treaty’s binary.
  • CTBT: not signed, with a voluntary unilateral moratorium on testing and an undertaking not to obstruct entry into force.
  • FMCT: supports a universal, non-discriminatory, internationally verifiable treaty, while producing fissile material.
  • TPNW: did not negotiate, did not sign and voted against the enabling resolution, on the ground that the process bypassed the Conference on Disarmament, excluded the nuclear-armed states, and does not meet the verification and universality requirements India insists on.
  • Disarmament: continues to table General Assembly resolutions on a convention prohibiting the use of nuclear weapons and on reducing nuclear danger, including measures against unintentional and accidental use.

India’s argument with the nuclear order has never been that non-proliferation is wrong; it is that a rule dividing states permanently by a date in 1967 is not a rule but a settlement of accounts.

Liability, energy and the 2025 recasting

• The civil nuclear deal did not produce reactors, and the obstacle was domestic law.
  • The Civil Liability for Nuclear Damage Act, 2010 imposed strict, no-fault liability on the operator, capped at ₹1,500 crore, with the government liable above that up to the rupee equivalent of 300 million Special Drawing Rights.
  • Section 17(b) gave the operator a right of recourse against the supplier where an incident resulted from equipment or material with patent or latent defects or sub-standard services — a departure from the international principle of exclusive operator liability, adopted because defective components had featured in the Bhopal disaster.
  • Section 46 was read as leaving suppliers exposed to claims under other civil laws and to criminal liability, and therefore to potentially unlimited exposure.
• The commercial consequence was two decades of paralysis, with no foreign-supplied reactor built under the deal.
  • Jaitapur with France and Kovvada with the United States stalled, and the only foreign presence remained Russia at Kudankulam, which predated the Act.
  • India ratified the Convention on Supplementary Compensation in 2016 and argued that Section 17(b) permits but does not require recourse, without reassuring suppliers or their insurers.
• The position changed decisively with legislation passed in December 2025 — the Sustainable Harnessing and Advancement of Nuclear Energy for Transforming India (SHANTI) Act, which repeals both the Atomic Energy Act, 1962 and the Civil Liability for Nuclear Damage Act, 2010.
  • It opens the sector to private and foreign participation in generation, reactor supply and operation, ending the state monopoly held since 1948.
  • It restructures liability, tying the operator’s cap to installed capacity and weakening the supplier-recourse provisions that deterred vendors.
  • It gives the Atomic Energy Regulatory Board statutory status, which it never had.
  • The criticisms are that the caps are low against the scale of a serious accident, that regulatory independence remains incomplete because appointments stay with the government, and that limitation periods constrain victims’ claims.
• On the supply side, American approval in 2025 for Holtec to design and build small modular reactors in India with Indian industrial partners was the first specific authorisation of its kind, and India’s nuclear expansion target for 2047 places small modular reactors and private capital at its centre.

Where the Regime Stands

• The bilateral architecture has ceased to exist. New START expired on 5 February 2026 with nothing to replace it, no verification, and an unresolved argument over whether a successor must include China.
• The review process has failed a third consecutive time. The Eleventh Review Conference closed in May 2026 without a substantive outcome, deadlocked over Iran, North Korea and Ukraine’s nuclear facilities.
• The testing moratorium is under strain. Russia de-ratified the CTBT in 2023; the American administration announced in October 2025 that it would resume “testing”, with the meaning unresolved. No state other than North Korea has conducted a nuclear explosive test this century.
• Iran’s programme is unverified. After the June 2025 strikes, the July 2025 suspension of cooperation and the reimposition of UN sanctions on 28 September 2025, the IAEA cannot account for the 60%-enriched stockpile or confirm whether enrichment continues.
• North Korea has consolidated: roughly fifty to sixty warheads, constitutionally entrenched nuclear status since 2023, a legislated use doctrine since 2022, and no diplomatic process.
• Arsenals are growing again. SIPRI counts about 12,200 warheads worldwide in January 2026, with China’s growth the steepest, all nine possessors modernising, and the global total rising for the first time in decades.
• The prohibition treaty is consolidating without the possessors: around 75 states parties, three Meetings of States Parties completed, and the First Review Conference set for New York in late 2026.
• Proliferation pressure is higher than at any time since the 1960s — alliance doubt in Seoul, Tokyo and Warsaw, an American openness to allied fuel-cycle capabilities reversing thirty years of practice, and the Ukrainian precedent.

Conclusion

The nuclear order has always been a bargain rather than a prohibition, and bargains fail when one party stops performing. What is happening now is not a sudden crisis but the accumulated cost of five decades in which the disarmament pillar carried no weight. The achievement remains extraordinary — nine nuclear states rather than the twenty-five once forecast, eight decades of non-use, and a norm strong enough that acquisition still carries a price. But that achievement rested on denial, deterrence and legitimacy together, and it is the legitimacy that is now thinnest.