CISS Strategic View – From PMDA to Damage Limitation: How a Broken Materials Regime Supercharges the Offence–Defence Spiral

Why a 25-year-old plutonium accord still matters

On 8 October 2025, Russia’s State Duma voted to withdraw from the U.S.–Russia Plutonium Management and Disposition Agreement (PMDA), the 2000 accord that committed each side to render 34 metric tons of weapons-grade plutonium (WG Pu) unusable for warheads. This 34 metric tons of WG Pu is roughly the fissile equivalent of ~17,000 nuclear weapons) (Reuters; State Department). Some American experts noted when Moscow suspended the PMDA in 2016, the dispute was never purely technical; it was a signal that the last functioning channel of U.S.–Russian materials verification was becoming leverage in a larger confrontation (ACT). The 2025 withdrawal formalizes that shift. In a year when the CTBT norm has come under visible stress and nuclear signalling has intensified, losing PMDA doesn’t just reduce transparency around plutonium. It accelerates the drift from verification-based stability to demonstration-based deterrence. 

 What PMDA was meant to do—and why it broke

Signed in 2000 and updated by protocol in 2010, PMDA aimed at irreversibility: each side would dispose of 34 tons of WG Pu under monitored, technically credible pathways. Initially, the U.S. did this via mixed-oxide (MOX) fuel; Russia via MOX in light-water and fast reactors (U.S. State Department). 

The U.S.’ MOX fuel project at Savannah River reportedly became a classic cost-and-schedule failure. After roughly $6 billion had been spent with persistent overruns, the Department of Energy’s (DoE) National Nuclear Security Administration (NNSA) cancelled the project in October 2018. Washington then pivoted to dilute-and-dispose (D2D): down-blending surplus WG Pu and emplacing it as contact-handled transuranic (CH-TRU) waste in the Waste Isolation Pilot Plant (WIPP) near Carlsbad, New Mexico.

Initially, D2D covered 6 to 7.1 metric tons under decisions taken in 2016 and 2020. In 2024, DoE’s program-level Record of Decision identified D2D as the preferred method for disposing of up to 34 metric tons (GAO-19-25; Federal Register 2020 AROD; DOE/Energy AROD; DOE/Summary). Moscow argued that the shift undermined the PMDA’s standard of irreversibility and labeled it non-compliant; first citing it to suspend participation in 2016 and again in 2025 to justify full withdrawal (ACT; Reuters; IPFM/Princeton).

Therefore, PMDA’s collapse is not merely about a factory that failed. It is about the politicization of verification itself: when an adversarial relationship hardens, even the most technical transparency regime becomes a bargaining chip. 

The larger scaffolding

PMDA’s demise lands in a system where classic arms-control scaffolding has already thinned. In June 2002, the U.S. withdrew from the Anti-Ballistic Missiles (ABM) Treaty with Russia, opening the door to unconstrained homeland and regional defences (White House; Arms Control Association).  The Intermediate Nuclear Forces (INF) Treaty formally ended on 2 August 2019 after mutual non-compliance accusations by Moscow and Washington (Arms Control Association; ACT news).  The U.S. withdrew from the Open Skies Treaty in 2020, forcing Russia to end a key overflight transparency channel (U.S. State Department). 

On 21 February 2023, Russia suspended participation in New START, halting inspections and routine data exchanges while saying it would continue to observe the Treaty’s central numerical limits and maintain launch notifications. On 22 September 2025, President Putin proposed that Russia would voluntarily keep limits for one year beyond the Treaty’s 5 February 2026 expiry, if the U.S. did the same (Reuters; AP). 

In October 2023 Moscow also de-ratified the CTBT to “mirror” Washington’s long-standing non-ratification; and in October 2025, U.S. political signals around nuclear testing further strained the taboo (Reuters; CTBT de-ratification; Reuters; Reuters follow-up). 

As the arms control guardrails painstakingly built during the ‘Cold War’ have buckled over the past two decades, the risk of ‘Hot War’ has grown. Where rules thin, signals fill the vaccum: lone ‘no’ votes and pointed ‘abstentions’ – even in non-binding UN General Assembly resolutions – ambiguous directives to resume nuclear testing, conspicuous exercises, and choreographed unvelings of new systems for which there is no ready antidote. Crises are managed less by procedures than by performance. The PMDA’s disappearance deepens this shift, removing a backstage mechanism that once translated political vows into verifiable acts. The choice is stark: pursue mirage of invulnerability through armament, or invest in deterrence grounded in mutual vulnerability.

 Threshold compression

The PMDA’s loss coincides with three accelerants that compress warning-to-decision time. First, hypersonic cruise and glide systems and manoeuvrable re-entry vehicles shrink detection and interception windows while blurring payload ambiguity i.e. intensifying ‘use-or-lose’ pressures (CISS AJK; Carnegie).  Second, introducing AI into early warning and nuclear, command, control and communications (NC3) speeds processing but raises risks of automated misperception and brittle human-machine teaming (CISS AJK; CNAS; CNA). Third, space–cyber entanglement enables non-kinetic interference with dual-use satellites—jamming, spoofing, or cyber intrusion—that can resemble pre-attack preparation and steepen escalation ladders (Secure World Foundation 2022/2024; 2024 update). 

When verification drains from the system, speed replaces it. That is the deeper strategic cost of losing a materials-accounting regime like PMDA, as it removes friction exactly where new technologies reward haste. 

India–Pakistan transmission belt

Arms-control erosion at the core radiates to regions with short warning times and thin political buffers. South Asia is the canonical case. Two surviving India–Pakistan arrangements i.e. the 1988 Non-Attack Agreement on nuclear facilities with its annual January 1 exchange of lists and the 2005 agreement on pre-notification of ballistic-missile flight tests, remain valuable but were designed for a slower era (NTI; AP; MEA; ACT). With hypersonics, cruise systems, algorithmic warning, and space-reliant intelligence-surveillance-reconnaissance (ISR), those instruments no longer cover the most ambiguous and rapid ladders of escalation. If PMDA’s withdrawal signals that great-power verification is optional, regional hedging becomes easier; and the persistent Indian debate about thermonuclear validation gains geopolitical cover when testing talk resurges among the majors (Jacob). 

Designed deceleration

If formal arms control cannot quickly return, the task is to re-insert time and procedures into systems optimized for speed. Three layers of practical steps would help i.e. P5+ risk-reduction track, replacing PMDA with PRVI, and South Asia CBMs 2.0.

P5+ risk-reduction track

Extend the logic of the 2005 Pakistan–India ballistic pre-notification regime to include hypersonics and cruise missiles, using UNODA as a shared notice hub by building on existing bilateral templates. 

Create a P5+ forum (the five NPT nuclear-weapon states plus other non-NPT nuclear powers) on AI in nuclear warning and command. They can meet biennially to share a short, non-sensitive brief explaining where AI or machine learning is used in early warning and decision support, which functions are strictly off-limits (for example, target selection or launch authorization), who retains authority to halt or override machine outputs, and what fail-safes and rollback procedures exist after false alarms. Adopt a common glossary, publish a concise public summary of agreed principles, and hold a closed-door technical session for lessons learned. The aim would be to clarify how automation is, and is not, used in nuclear systems, and reduce the risk that can accelerate a crisis (CNAS).

Adopt a ‘no-blind’ pledge i.e. in any crisis, all parties commit not to degrade early-warning and NC3 satellites by jamming, spoofing, dazzling, or cyber interference. Specify which assets are covered (missile-warning, strategic communications, and ISR satellites) and which actions are prohibited, while allowing routine spectrum management and safety operations. Pair this with a no-fault incident reporting scheme, modeled on aviation and maritime safety programs: operators report anomalies confidentially to a neutral hub, facts are shared quickly without attribution, and fixes are circulated as lessons learned. The aim is simple i.e. to keep eyes and ears open during crises so misreads do not trigger escalation (SWF counterspace reports 2022/2024).

Replace PMDA’s function with Plutonium Re-Verification Initiative (PRVI)

Host IAEA-led technical audits of civilian plutonium that cover inventories and key steps such as assay, packaging, and movements—with voluntary peer review among states that opt in. On the U.S. side, publish a consolidated dilute-and-dispose (D2D) schedule and allow limited observation at defined milestones to show that D2D is real and traceable (Federal Register 2020 AROD; DOE 2024 ROD). On the Russian side, if withdrawal proceeds, pair it with a moratorium on reconversion or re-weaponization and greater transparency over BN-800/BN-1200 fuel flows. Together, these steps would demonstrate practical irreversibility even without the original legal framework (Reuters). Russia’s BN-800 (operational since 2016) and planned BN-1200 and BN-1200M are sodium-cooled fast-neutron reactors at the Beloyarsk plant designed to run on MOX fuel (Rosatom), with BN-1200 also envisaging nitride fuel in a closed cycle (IAEA, WNA). Because these units can “burn” weapons-grade plutonium in power-reactor fuel, their fuel flows are central to demonstrating practical irreversibility and to any transparency regime that might substitute for the defunct PMDA (ENS).

South Asia CBMs 2.0

Dialogue between India and Pakistan has been largely frozen since 2012, with New Delhi tying any resumption to demonstrable progress on countering terrorism. Islamabad has signalled readiness for talks that treat terrorism as a shared challenge. Recurrent crises since then—including pre-emptive Indian strikes, rapid military mobilisation, and Pakistan’s defensive responses to restore conventional deterrence in a nuclearised environment—have further eroded trust. Rebuilding confidence is therefore essential, not optional.

A practical first step is a No-Blind Pledge in which both sides commit not to interfere with each other’s early-warning and NC3 networks—no jamming, spoofing, dazzling of satellites, or cyber intrusion—especially during crises. Protecting ‘eyes and ears’ reduces the risk of misreads, false alarms, and inadvertent escalation.

Why PMDA’s loss bites harder now

In October 2025, the nuclear signalling environment was unusually dense: NATO’s Steadfast Noon drill and U.S. Strategic Command’s Global Thunder 26 ran alongside Russia’s publicised Burevestnik flight and a U.S. presidential directive to restart nuclear weapons testing—without clearly ruling out underground shots (NATO/ACO; USSTRATCOM; Reuters; Trump directive; Reuters; CTBTO). Against that backdrop, losing the PMDA is not merely the end of a technical program; it removes one of the few materials-accountability brakes left in the system, letting a spiral powered by speed, ambiguity, and public performance spin faster.

A Policy Menu

For Washington. Treat dilute-and-dispose (D2D) as a credibility programme, not merely a budget line. Set and publish time-phased milestones, invite independent validation, and allow limited observation at defined points to demonstrate irreversible down-blending without MOX. In parallel, strengthen NNSA programme controls in line with GAO guidance  (GAO-19-25) to avoid a repeat of the Savannah River governance and cost-overrun problems.

For Moscow. If legal symmetry with the United States is the aim, pair withdrawal with practical constraints: adopt a codified moratorium on reconversion or re-weaponisation of separated plutonium; provide regular transparency on BN-800/ BN-1200 fuel fabrication, loading, and discharge; and consider IAEA-notified data releases or peer review at agreed milestones. This preserves bargaining leverage without inviting the narrative of unconstrained re-armament (Reuters, Oct. 2025). 

For Vienna-based institutions and willing member states. Stand up a pilot Plutonium Re-Verification Initiative (PRVI) under the IAEA – with support from the CTBTO, donor governments, and technical labs – to keep verification skills, datasets, and cooperative habits alive. Start with a few volunteering countries, run time-boxed audits and peer reviews, publish anonymised methods, and fund it through voluntary contributions. In an era of fast code and hypersonic flight, materials transparency is the slow, steady buffer the system can still build.

For South Asia. Pair the ‘No-Blind Pledge’ with a light-lift practice i.e. a no-fault anomaly bulletin, which would be a short, standardised notice within an hour when either side experiences a space/ NC3 disruption that gives time, asset class, symptom, expected duration, but no attribution. Both measures protect the same ‘eyes and ears’ without reopening contentious agendas, and they cut the risk that technical glitches are misread as hostile acts

Conclusion

PMDA’s end is a warning about tempo. As verification recedes, crisis management defaults to performance at algorithmic speed. The middle ground is not wishful trust but relying on the re-worded Russian maxim: ‘trust and verify.’ In this decade, stability will depend less on grand treaties than on friction: mechanisms that slow kill chains, safeguard early-warning, and widen decision windows. If we cannot yet agree on how far to go, we should at least agree on how fast. Re-creating time is the most practical way to keep the offence–defence spiral from ending in catastrophe.