Explosive nuclear testing resumes and the 2031 review conference ends the numerical era without a successor
At least one recognised nuclear-weapon state conducts an explosive nuclear test detected and radionuclide-corroborated by the international monitoring…
Claude · 2031 · grey-swan
Prior state
Verified numerical limits on the two largest strategic arsenals lapsed in February 2026 with no successor and no negotiating framework. The third large arsenal continued a declared expansion. Site-readiness activity was publicly discussed by more than one nuclear-weapon state during 2025 and 2026. Modernisation programmes in all five recognised nuclear-weapon states moved into production of designs that were certified, but never proof-tested, under the moratorium regime — creating a standing technical argument inside each weapons establishment for validation. The comprehensive test-ban treaty remained unratified by several key states and therefore in force only as a political norm underwritten by an operational monitoring network.
Material change
At least one recognised nuclear-weapon state conducts an explosive nuclear test detected and radionuclide-corroborated by the international monitoring system, ending a moratorium held since the mid-to-late 1990s. At least one other announces readiness or follows. The 2031 review conference ends without a consensus final document and without any successor framework containing numerical limits, and hedging states in Northeast Asia, the Gulf, and Central Europe move from academic debate to formal governmental capability review. The material change is the loss of the two constraints that survived the collapse of the treaty architecture: the testing taboo and the presumption against overt latency.
Why now
The review conference falls in 2031 under the five-year cycle and is the scheduled occasion on which the non-nuclear-weapon-state majority must formally judge whether the bargain has failed. It is also five years since the last numerical instrument expired, which is long enough for the modernisation programmes begun in the 2010s to reach the production and certification stages where the technical case for testing becomes concrete, and long enough for the political cost of being first to test to have been absorbed into normal discourse. A test in 2027 would have preceded that technical demand; a test after 2031 would follow a review conference that had already been allowed to fail.
Mechanism and resistance
The mechanism is reciprocal and fast: a first detected test removes the principal cost of a second, and the states whose programmes have the strongest technical case are also those with intact test infrastructure. Resistance is substantial, which is why this is a low-probability branch rather than a default. The monitoring network makes concealment impractical for any yield of military interest, so a test is a deliberate political act rather than a covert one. Weapons laboratories in several states genuinely prefer subcritical experiments and computational certification, having built their careers on them. Allied governments impose real costs. And a state that tests hands its rivals a justification it may not want to grant. The path runs through a specific decision by a specific leadership under perceived pressure, not through drift.
Consequences
The immediate consequence is to convert nuclear latency from a hedge into a policy option that governments can discuss publicly, which is the mechanism by which proliferation decisions have historically been made. The states most affected are those with advanced civil nuclear infrastructure, a security dependency they have come to doubt, and a domestic political tradition that has kept the question suppressed. A second consequence is environmental and jurisdictional: the states and territories that hosted twentieth-century testing, and the Pacific and Central Asian governments that led the ban's diplomacy, become the moral centre of a resistance campaign with real convening power but little leverage. A third is arms-racing dynamics without the transparency mechanisms that made the twentieth-century races calculable — no on-site inspection, no declared deployment counts, and estimation left to open-source analysis.
End state
A nuclear order with no verified numerical limits, a broken testing norm, a review process that has failed publicly, and at least two additional states with formal governmental work programmes on independent capability — the least constrained configuration since the early 1960s.
Observable test
A monitoring-system-detected explosive nuclear event with radionuclide corroboration attributed to a recognised nuclear-weapon state; the final-document status of the 2031 review conference; and the existence of publicly acknowledged governmental capability reviews in any non-nuclear-weapon state party.
Disconfirming sign
No detected explosive test by any recognised nuclear-weapon state through 2031, and the review conference adopts a consensus final document reaffirming the moratorium.