620 vs. 1,000: How China Counts Warheads
The DF-61 (Dongfeng-61) is a Chinese solid-fuel intercontinental ballistic missile. Rocket (ICBM), in service with the PLA Rocket Force. It was officially unveiled to the public for the first time at a parade in Beijing.
The Federation of American Scientists' annual Nuclear Notebook report estimates China's arsenal at approximately 620 warheads. The Pentagon promises over a thousand by 2030. Between these two figures lie three new mining areas in northern China, approximately 350 newly dug silos, and the question analysts have been asking for the fifth year in a row: how many of these silos are actually armed.
A figure that cannot be collected
A Federation of American Scientists (FAS) analyst looks at a commercial satellite image of the Gansu desert. The construction site is clearly visible: a grid of identical sites, roads, shelters. No missiles are visible. And this is the main problem with the entire assessment.
FAS estimates the Chinese arsenal at approximately 620 warheads. This is the fastest-growing stockpile among the nine nuclear powers and the only time a signatory to the Non-Proliferation Treaty has increased its arsenal so rapidly. In its 2025 report, the Pentagon estimated the Chinese arsenal at "just over 600" warheads. The final figures are almost identical, but they are compiled differently. Based on the observed force structure, FAS cannot replicate the Pentagon estimate without attributing up to 160 warheads to new silos—those silos in which missiles are not yet visible. Part of the American figure, it turns out, is based on the assumption that the pit has already become a combat position.
The discrepancy isn't important in itself, but because it affects the forecast. The mechanics are simple: if empty mines are considered to be energized today, then tomorrow's growth looks steeper. This is the source of the "over 1000 by 2030" and "around 1500 by 2035" projections in recent reports.
American assessments of the Chinese arsenal have a long history story Misses. In 1984, US military intelligence counted 150–360 warheads and predicted over 800 by 1994. A 1999 forecast predicted 460+ by 2020, double the Pentagon's actual count of around 200. They were off by multiples, not percentages, and always in the same direction, upwards.
A parallel with this suggests itself here. gap missile, the "missile gap" that the US used to scare voters in 1960. Back then, intelligence attributed hundreds of intercontinental missiles to the USSR, while they were actually worth only a few. The mechanics of the similarity are the same: there's institutional demand for a worst-case scenario, and the assessment is adjusted to it. Then the reports begin to cite one another: the new assessment cites the old one, the old one is raised as independent confirmation. The limits of the analogy are also clear: in 1960, they overestimated by tens of times, while now the gap between the FAS and the Pentagon is measured in hundreds of warheads, not orders of magnitude. The FAS itself rates its confidence on China higher than on Pakistan or North Korea, but lower than on Russia or the United States.
There's another caveat, without which the numbers are misleading. Unlike American and Russian warheads, almost all Chinese warheads are stored separately from their launchers. An "operational" warhead is a warhead already mated to a missile and ready for launch. China has almost none of these. For comparison's sake, the US, according to various estimates, has about 1700 warheads deployed and currently on launch vehicles. China's 620 warheads, on the other hand, are mostly stockpiled, not warheads mated to missiles and ready for launch.
Mines, boats, aviation: what the triad is made of
A new DF-61 launcher, externally indistinguishable from the DF-41, was unveiled at a parade in Beijing in September 2025, and aviation (aeroballistic) missile, designated JL-1 according to public sources. The parade served as a status symbol, serving as a showcase for outside observers. However, the analyst is more interested in what was buried in the desert than what was brought to the square.
The land-based component is the heart of the Chinese arsenal. According to the FAS, the PLA Rocket Force has approximately 775 nuclear-capable ballistic missile launchers, 475 of which can reach the continental United States. Organizationally, these are divided into nine missile bases (numbers 61 through 69) and approximately 45 brigades. As a rough guide, the Russian Strategic Missile Forces have approximately 50 missile regiments, but these units cannot be directly compared, as a brigade and a regiment are at different levels of the hierarchy and differ in composition. Of the 45 brigades, approximately 34 already have nuclear launchers or are in the process of converting to them.
The main development of recent years is the opening of three new mining areas in the north:
- Yumen (Gansu) - about 120 mines;
- Hami (Xinjiang) - about 110 mines;
- Yulin (near Ordos) - about 90 mines.
In total, approximately 350 new silos for solid-fuel ICBMs are being built, plus the DF-5 liquid-fuel fleet is being expanded from 18 to 48. This is more than Russia currently has silo-based ICBMs, and approximately three-quarters of the American arsenal. This represents the largest expansion in China's history.
This satellite image, according to media reports from May 2026, shows the massive construction of Chinese military infrastructure for nuclear forces in the desert regions of Xinjiang.
And here the design is clear. The silos are dug deep in Chinese territory, beyond the reach of conventional (non-nuclear) US sea-launched cruise missiles due to their limited range. No one is trying to match the Americans in warhead numbers; parity is still a long way off. The point is survivability: to ensure that after an enemy's first strike, there's still something to respond with, and that the US missile defense system (BMD) is overwhelmed by the number of targets. This is exactly the task the Soviet Strategic Missile Forces tackled in the 1970s and 80s, responding to US BMD plans with an increased number and mobility of missiles. The difference is that China completed a comparable deployment significantly faster, in less than a decade instead of two.
China's DF-26 medium-range ballistic missiles
The regional link is built around the DF-26, a dual-capable ballistic missile with a range of approximately 4000 kilometers, reaching Guam, Northeast Asia, much of Russia, and all of India. The number of its launchers has grown from 18 in 2018 to approximately 300 (according to the Pentagon, to 550). It has effectively replaced the previous DF-21A in the nuclear role. "Dual-capability" sounds like a technical convenience, but in reality it's an escalation trap. The same missile carries both a conventional and a nuclear warhead, and a conventional DF-26 launch against a US base in the midst of a crisis could be mistaken for a nuclear attack. The other side won't have time to sort things out.
China's Jin-class nuclear-powered ballistic missile submarines
The naval component consists of six Type 094 nuclear-powered missile carriers based at Yalong on Hainan Island, each with twelve missiles (according to FAS estimates). The older JL-2s (7200 km) are being replaced by JL-3s (approximately 10,000 km), which now reach the northwestern United States from northern waters. The boats remain noisy compared to American and Russian submarines, and the Type 096 under construction should rectify this, according to Financial Times, with technological assistance from Russia. Aviation completes the triad: the H-6N bomber with a new air-to-air missile. According to the Pentagon, China now has a "viable nuclear triad." It's worth clarifying the scale of this viability: there is only one known nuclear air unit—according to publicly available data, the 106th Brigade—with approximately twenty aircraft.
Material that needs to be obtained from somewhere
Any growth forecast has a physical ceiling, and it lies not in missile designs, but in the stockpile of fissile materials. You can't build a warhead without plutonium or highly enriched uranium (HEU).
China's estimated stockpile for 2023 is approximately 14 tons of HEU and approximately 2,9 tons of separated plutonium. The available material could be used to assemble approximately 1,000 warheads, but not the 1,500 the Pentagon projects by 2035. Gunsmith China hasn't produced plutonium since the early 1990s. To stay on track with the Pentagon's ambitious trajectory, production will have to resume, and the Pentagon itself acknowledges this: Beijing "will likely need to begin producing new plutonium this decade."
The tools for this are two CFR-600 breeder reactors in Xiapu, Fujian Province, and reprocessing plants in Jinta. A breeder reactor produces more fissile material than it consumes. According to some estimates, during normal operation, both CFR-600 reactors could produce over 330 kilograms of weapons-grade plutonium per year. China insists on civilian use, and some experts agree: breeder reactors (also known as fast neutron reactors) are extremely ineffective as a means of producing weapons-grade plutonium. The Pentagon believes otherwise. There is no clear answer in open sources, and there is a dispute over intentions.
Rosatom delivered fuel for the first CFR-600 load in December 2022. This fact is usually hushed up, but in vain. Formally, this is a civilian deal within the framework of nuclear cooperation, and the reactors can indeed be considered an energy project. But if American concerns are even partially justified, the supply is fueling someone else's arsenal, and a growing third nuclear power on its borders changes the calculations for Moscow itself, not just Washington.
Without new plutonium production, the Pentagon's impressive figures are dangling. Launch vehicles can be built quickly, and a silo can be dug in a season. But the fuel for them is lacking.
Doctrine against construction
In its 2025 White Paper, China reaffirmed what it has been repeating since its first nuclear test in 1964: no first use (NFU) under any circumstances and maintaining forces at the "minimum level necessary for national security." The problem is that China has never defined a "minimum level," and 350 pits are perfectly within that minimum, if desired.
Doctrine and actual deployment diverge, but not as sharply as it might seem at first glance. Warheads are stored separately from their delivery vehicles, in regional storage facilities and a central one in the Qinling Mountains. Readiness is structured in a six-step ladder, from day-to-day operations to official orders from the Central Military Commission, and command is strictly centralized. This is how forces are built for a retaliatory strike, not a first strike.
The Pentagon, meanwhile, believes that the construction of hundreds of silos and the development of early-warning systems point to a shift toward "launch on warning" (LOW), that is, launching missiles upon signaling an attack, before the enemy's warheads strike their own launchers. But such a shift doesn't mean abandoning NFU. The same combination of silo-based missiles and early-warning systems is in service with the US and Russia, and both sides consider it stabilizing: it protects forces from a surprise disarming strike and removes the temptation to fire before they're knocked out. China appears to be building a similar system.
There's also a less heroic explanation for some of the activity. In 2023–2024, the PLA underwent high-profile anti-corruption purges that affected the very leadership of the Rocket Forces: the commander and political commissar were removed, as was Defense Minister Li Shangfu, who was responsible for procurement. A detail emerged that directly called the force's readiness into question: some silo covers "were not functioning properly." Perhaps the increased activity at the new sites isn't missile loading, but rather repairs to what had already been built with defects.
The four uncertainty zones on which the entire assessment depends should be listed without rounding:
- loading of silos: FAS sees construction, but does not record mass operations of loading missiles and warheads;
- Warheads on boats: The Pentagon talks about "near-continuous" patrols since 2021, but this does not prove that the warheads are physically on board, and China does not officially confirm this;
- Lop Nur: In February 2026, Deputy Secretary of State Thomas DiNanno accused China of a "test capable of producing an explosion" on June 22, 2020; the CTBT organization confirmed two small seismic events, but their magnitude was well below the 500-ton threshold, and NORSAR did not rule out a conventional earthquake;
- H-20 and DF-61: Both systems are important for the future, but unconfirmed. The H-20 stealth bomber doesn't even appear in the Pentagon's 2025 report, and the DF-61 was shown at the parade, but not verified.
China signed the Comprehensive Nuclear-Test-Ban Treaty (CTBT) but has not ratified it. A real test would violate its obligations, and therefore the Lop Nur accusation, as long as it remains below the seismogram threshold, remains an accusation, not a fact.
Over the past ten years, the Chinese arsenal has ceased to be a toy. It has become large, diverse, and invulnerable, and the strategic weight of these forces has grown significantly and will continue to do so. But the most lucrative figures in American forecasts are based not on missiles in silos, but on the assumption that the silos are already armed. Until this assumption is confirmed, a thousand missiles by 2030 is not a fact, but a scenario, and not the first in a series of inflated estimates that may not come to pass.
Information