China’s Nuclear Buildup Spans 7 SECRET Sites

Missile underside with Chinese flag overlay — Photo: Allexxandar / Shutterstock

China’s nuclear forces are growing in size and complexity, and the strongest public evidence comes from independent stockpile estimates and repeat satellite imagery of new missile silos and weapons-related infrastructure—not from official Chinese disclosures.

At a Glance

SIPRI estimates China’s stockpile reached roughly 620 warheads by January 2026 and is expected to keep rising.
Commercial imagery and open-source analysis have documented three large new ICBM silo fields and additional silos under construction in China’s east.
A media investigation highlights seven sites—ranging from pit production and warhead R&D to testing and fuel-cycle work—consistent with a broad nuclear-industrial scale-up.
Beijing publicly rejects “arms race” claims, reiterating a no-first-use doctrine and minimum deterrent posture.

What the hard evidence actually shows

The most reliable public indicator of China’s current nuclear trajectory is the annual assessment by the Stockholm International Peace Research Institute (SIPRI). Its 2026 Yearbook places China’s arsenal at approximately 620 warheads—up from around 600 the prior year—and judges that growth will continue. SIPRI’s reporting also aligns with the surge in visible infrastructure: by 2025–2026, analysts had identified hundreds of new intercontinental ballistic missile (ICBM) silos across three large fields in China’s north, with further silo construction in mountainous terrain to the east—features consistent with a force designed to be more numerous, more survivable, and more responsive.

Those headline numbers do not stand alone. Civil-society imagery analysts first mapped the large silo fields at Yumen, Hami, and Ordos beginning in 2021, tallying roughly 120, 110, and 40 silos respectively, together with roads, berms, and support facilities that match known ICBM operating patterns. The open record captures a sustained, multi-year build-out rather than a one-off site that could be misread. In the world of nuclear force structure, concrete and dirt tell their own story.

The seven-site picture: weapons complex or mosaic of dual-use projects?

A recent investigation synthesizes commercial imagery and specialist interviews to name seven locations—Pingtong, Zitong, Lop Nur, Jiuquan, Mianyang, Jinta, and Guangyuan—where construction and configuration are deemed consistent with nuclear weapons–related functions. The claim is not that each site is exclusively military, but that the ensemble signals a scaled industrial base: reported pit-production capacity and underground storage at Pingtong; laser-driven physics at Mianyang of the type used globally to model warhead behavior; diagnostic and initiation testing at Zitong; ongoing tunnel and support work at the Lop Nur test complex; and fuel-cycle and waste-handling expansion at Jiuquan. None of these functions proves a warhead on the end of a missile tomorrow. Taken together, they do indicate an ecosystem maturing in parallel with the silo expansion.

Two caveats belong alongside those details. First, the seven-site narrative relies on imagery interpretation and expert inference; it is not underwritten in public by Chinese program documents connecting every dot. Second, several facilities are plausibly dual-use: high-energy lasers, uranium conversion, and materials R&D serve civilian and military ends. These are reasons to weigh the claims carefully, not to dismiss the broader conclusion that China is fielding more delivery capacity and the infrastructure to support it. On that point, the silo fields and SIPRI’s stockpile accounting carry most of the evidentiary weight.

China’s stated doctrine and how to read it against the build-out

Beijing’s official line is consistent across ministries and venues: no-first-use under any circumstances, negative security assurances for non-nuclear states, opposition to proliferation, and a pledge to avoid an arms race. Chinese statements cast current activity as modernization for reliability and safety within a minimum-deterrence framework. Whether one accepts the doctrinal purity of no-first-use, the declaratory policy is not a trivial talking point; it shapes force design. A strict NFU state must ensure a survivable second strike in the face of improving adversary defenses and conventional strike capabilities. That logic supports hardening, dispersal, decoys, mobility, and numbers sufficient to ride out a first blow and still retaliate.

From that lens, large silo fields and diversified R&D can be read as an attempt to move from a historically “thin” deterrent toward a posture resilient against counterforce and missile defense. The empirical question is scale and intent: does a shift from hundreds to potentially many hundreds more warheads and the infrastructure to sustain them remain “minimum,” or does it mark a structural breakout? Public data can illuminate growth and capability; it cannot, on its own, adjudicate motive. Here, SIPRI’s yearbooks provide the quantitative spine; the seven-site thesis sketches the qualitative edges.

Mechanisms and milestones: how an arsenal expands

Building warheads is only one gate in a chain. Industrially, a modern program must produce or reprocess special nuclear material, fabricate pits and other components to tight tolerances, complete non-nuclear explosive packages and neutron initiators, and validate designs through hydrodynamic tests and high-energy-density physics. Operationally, forces require delivery systems, basing modes, command and control hardened against cyber and kinetic attack, and trained crews. The observed Chinese pattern—fuel-cycle growth at Jiuquan, reported pit work at Pingtong, laser and diagnostics capacity at Mianyang and Zitong, tunnel activity at Lop Nur, and massed basing for missiles—tracks that manufacturing-to-fielding logic, even if individual nodes remain dual-use or partly opaque.

Missile silos, in particular, are a strategic signal with technical purpose. In the U.S.-Soviet experience, large fields complicated adversary targeting, supported shell-game deception with empty silos, and reduced the unit cost of security and infrastructure. China’s new fields mirror those incentives. Pair them with mobile ICBMs and sea-based assets, and you approach a triad whose redundancy deters preemption—an intuitively consistent move for a state that claims NFU yet worries about adversary advances in precision strike and missile defense.

China keeps calling for the complete prohibition and destruction of all nuclear weapons, while secretly carrying out the fastest-growing nuclear buildup in the world.

This kind of deliberately misleading and false information is a pattern. I don’t see how any report on any topic…

— Guy Barnett (@GuyBarnett99830) September 23, 2026

Where the debate is real—and where it isn’t

The strongest points are not seriously in dispute in open sources: China’s stockpile has grown, and new missile infrastructure has multiplied. Those are anchored by SIPRI’s conservative estimates and repeated commercial imagery. The legitimate debate lies in the label and the line: is this primarily a modernization catch-up to secure a credible retaliatory force, or a covert, coordinated breakout premised on eventual numerical rivalry? Because much of the seven-site case rests on inference from dual-use signatures, reasonable analysts can disagree on exclusivity and intent. But the combination of more warheads, many more silos, and the accretion of supporting R&D and industrial capacity points to a durable upward slope.

Implications for strategy and arms control

Three consequences follow. First, crisis stability risks rise when force structures are in flux; unfamiliar capabilities and expanding target sets create misperception hazards. Second, verification gaps widen: without Chinese transparency, public debate defaults to satellite photos and external estimates, which are necessary but not sufficient to stabilize expectations. Third, any future arms control framework that includes China must grapple with counting rules for silos, mobile launchers, and warheads—exactly the architecture now coming online. For policymakers, the pragmatic task is to keep two ideas in tension: take seriously the empirical expansion documented by SIPRI and imagery, and test declaratory doctrine through verifiable proposals that would be cost-imposing only if no-first-use is insincere. Until then, analysis should center on what can be seen and counted—and on building guardrails for a deterrence geometry that is no longer static.