Recent comments by U.S. Government officials about secret Chinese nuclear testing have led to much speculation and interest. On February 6, 2026, Under Secretary of State for Arms Control and International Security Tom DiNanno revealed to the Conference on Disarmament in Geneva that 

“China has conducted nuclear explosive tests, including preparing for tests with designated yields in the hundreds of tons. The [People’s Liberation Army] sought to conceal testing by obfuscating the nuclear explosions because it recognized these tests violate test ban commitments. China has used decoupling – a method to decrease the effectiveness of seismic monitoring – to hide their activities from the world. China conducted one such yield producing nuclear test on June 22 of 2020.” 

On February 17, Assistant Secretary of State for Arms Control and Nonproliferation Christopher Yeaw clarified further that the event in question occurred “right near” China’s nuclear facility at Lop Nur in Xinjiang, and that it produced a seismic signal of magnitude 2.76 that was captured by a monitoring station in Kazakhstan. The specific yield of the test, Yeaw noted, was unclear, though he did say that the data suggested that it was “a 10 tons nuclear explosion – or 5 tons conventional equivalent.” (That 10-ton figure, Yeaw made clear, was calculated on the basis that the explosion was “fully coupled in hard rock” – which, as I’ll explain further below, it would not have been if it were a larger explosion that had been decoupled.) Yeaw also echoed DiNanno’s comment that China had been “preparing designated tests of hundreds of tons” in yield. 

How should the reader evaluate these claims? And what are the implications of these U.S. conclusions?

Evaluating the U.S. Claims

The Comprehensive Test Ban Treaty Organization (CTBTO) initially put out a statement that its monitoring system “did not detect any event consistent with the characteristics of a nuclear weapon test explosion at that time.” Nonetheless, that same statement also noted that the CTBTO’s system is only “capable of detecting nuclear test explosions with a yield equivalent to or greater than approximately 500 tonnes of TNT,” so it remains quite possible that the CTBTO would have missed an event of the limited size indicated by the U.S. revelations. And indeed it was subsequently reported that the CTBTO had in fact picked up “two very small seismic events, 12 seconds apart,” in the time period described by Yeaw, though these signals were too small to “assess the cause of these events with confidence.” So seismic data alone doesn’t seem to give us a lot of clarity.

Bear in mind, however, that the U.S. officials are clearly not relying just on seismic data. Recall, for instance, that both DiNanno and Yeaw noted that the Americans knew beforehand that China was “preparing” for tests of up to “hundreds of tons” of explosive yield – and note also that DiNanno used the term “tests” in the plural, rather than talking exclusively about a single test. This assessment therefore obviously draws upon intelligence information, which we shouldn’t be surprised they haven’t revealed to the public.

So what should we then, as observers on the outside, make of this? For my part, it’s worth noting that I have some degree of experience with these issues. Indeed, in my last period of government service I sat in the chair now occupied by DiNanno – at which point he reported to me – and I also served, like Yeaw does today, as the assistant secretary in charge of nonproliferation issues. Back in the George W. Bush Administration, moreover, I was the Deputy Assistant Secretary of State in charge of assessing arms control and nonproliferation compliance – a point at which Yeaw served as director of my Nuclear Affairs Office – and during the Clinton Administration I was a U.S. Senate staffer during the Senate’s famous vote to reject the Comprehensive Test Ban Treaty in October 1999. I was also an intelligence officer in the U.S. Navy Reserve from 1994 until 2011.

Drawing upon this background, I find the U.S. claims highly credible. 

To being with, it’s no secret that the “decoupling” DiNanno mentioned is a highly viable way to conceal a nuclear explosive test. This has been known for many years, and possible cheating on the CTBT through such decoupling techniques was in fact one of our major concerns during the Senate’s 1999 debates over CTBT.

As recounted in a 2002 study by the U.S. National Academy of Sciences (NAS), without special concealment techniques, “nuclear explosions with a yield of 1 kiloton (kt) or more can be detected and identified with high confidence in all environments” – and in hard rock in many parts of the world, down even to something like a tenth of a kiloton (i.e., 100 tons) of explosive yield. (Indeed, in some highly-studied areas such as the Russian nuclear test site at Novaya Zemlya, this detection threshold was said to be even lower: down to “0.01 kt (10 tons) or less.”)

That’s actually pretty good. But the report also noted that there exist “evasion scenarios that need to be taken seriously.” Among these is “cavity-decoupling” – that is, conducting a test in the middle of an otherwise empty chamber underground in order to reduce the extent to which the explosion transmits shock to the surrounding earth in ways that can be picked up elsewhere as a seismic signal. As the National Academy further explained in a 2012 report that basically reaffirmed its prior conclusions, a “fully decoupled” explosion would still produce some seismic waves in the Earth’s crust, but “most of the energy” would go into “increasing the gas pressure in the cavity, thereby reducing the apparent yield of the original explosion.” 

Another way help to reduce a test’s seismic signature might be to couple decoupling, as it were, with the use of a large steel containment vessel in which the test explosion would initially occur. As the Arms Control Association has reported, it has been known for years that the Soviet Union used to conduct some nuclear-weapons-related explosive experiments “in large ‘Kolba’ vessels designed to contain explosions with total energy yields (chemical explosive plus fission) up to 50 kilograms of TNT” without crossing the CTBT’s detection threshold. 

I’m pretty sure there’s no way to build a Kolba strong enough to contain an explosion of the size DiNanno and Yeaw have been talking about in connection with secret Chinese testing. One should remember, in this respect, that back in 1945 the United States built a colossal, 25-foot-long, 214-ton steel vessel nicknamed “Jumbo,” which it was initially hoped might contain the chemical explosive power of the Manhattan Project’s “Gadget” implosion device in the event that the “Trinity” test at Alamogordo, New Mexico, failed to produce nuclear fission yield. (As it turned out, Jumbo was never actually used.) Yet though Jumbo was apparently designed to contain a force of 50,000 pounds per square inch, even it probably would have fragmented from explosive yields of between 100 and 300 tons.

That said, I imagine using a large Kolba-type container might help at least reducethe seismic signature produced by a fully decoupled test, because it would force the explosive yield to break out of the steel container firstand only then to radiate outward through the excavated cavity to transmit energy to the surrounding earth. You surely couldn’t contain an explosion of “hundreds of tons,” but it stands to reason that a massive vessel could nonetheless make a decoupled test even stealthier. 

So how much difference might decoupling make? Well, the 2002 NAS report suggests it’s possible to achieve a “signal-reduction factor” of 70 for small tests. This has apparently been demonstrated in at least one historical test of around 0.4 kilotons – that is, about 400 tons. (It’s not clear whether that old example used a Kolba-type containment vessel, but if not, the effective signal reduction available to Chinese technicians today could perhaps be even greater.)

Ominously, the “hundreds of tons” figure cited recently by U.S. authorities as being the size of the tests China was preparing to conduct would seem to fall easily into this window of such “decouplable” testing.

U.S. Assessments, 2019-20

It’s also worth remembering that this is not the first time the United States has raised the issue of secret Chinese nuclear testing. In fact, this is a subject we took some pains to raise – and draw attention to – in the First Trump Administration. 

In May 2019, Lieutenant General Robert Ashley, the head of the U.S. Defense Intelligence Agency (DIA), announced at a Hudson Institute event with Hudson’s Rebeccah Heinrichs that 

“China continues to use explosive containment chambers at its nuclear test site, and China[‘s] leaders previously joined Russia in watering down the language of the P5 statement that would have affirmed a uniform understanding of zero-yield testing.” 

In a further statement issued on June 13, 2019, DIA said that China’s nuclear activities raised “questions about those activities in relation to the ‘zero yield’ nuclear weapons testing moratorium adhered to by the United States, the United Kingdom, and France” and were “inconsistent with the commitments undertaken by the United States, the United Kingdom, and France.” 

In the First Trump Administration, we followed up these revelations in our official 2020 report at the State Department on Adherence to and Compliance with Arms Control, Nonproliferation, and Disarmament Agreements and Commitments – a report, by the way, that Tom DiNanno and I were both involved in preparing. There, our formal compliance finding about China declared that 

“China maintained a high level of activity at its Lop Nur nuclear weapons test site throughout 2019. China’s possible preparation to operate its Lop Nur test site year-round, its use of explosive containment chambers, extensive excavation activities at Lop Nur, and lack of transparency on its nuclear testing activities – which has included frequently blocking the flow of data from its International Monitoring System (IMS) stations to the International Data Center operated by the Preparatory Commission for the Comprehensive Nuclear Test-Ban Treaty Organization – raise concerns regarding its adherence to the ‘zero yield’ standard adhered to by the United States, the United Kingdom, and France in their respective nuclear weapons testing moratoria.”

(You’ll also note, by the way, that this language – as well as Ashley’s earlier comments – include reference to “explosive containment chambers” – that is, “Kolbas.”)

Those sentences were all we had permission to say about China at the time. But remember that this was a report we published in June 2020 – which meant that it was drafted and worked through the painful U.S. interagency clearance process before the June 22, 2020, Chinese test described by DiNanno and Yeaw. Even then, however, we did flag at that point that “[a]dditional information” was provided at a “higher classification” level in our classified Annex to the report. I can’t tell you what was in that Annex about this Chinese nuclear testing question, of course, but we did make quite clear that we did have more information than we could publicly disclose at the time.

As you can see, therefore, some of the Chinese nuclear testing backstory has been available for several years, for anyone who cared to pay attention to it. This adds credibility to the current U.S. accounts.

The Implications

So what are the implications of these revelations about secret, yield-producing Chinese nuclear weapons tests? Most obviously, this is just another in a series of remarkable revelations – including the vast proliferation of Intercontinental Ballistic Missile (ICBM) silos in the deserts of Xinjiang and the expansion of nuclear weapons “pit” production at Pingtong– about the extraordinary pace of Beijing’s provocative and destabilizing nuclear weapons buildup. 

That buildup, in fact, seems well on track to give China strategic parity in terms of operationally deployed strategic warheads with both the United States and Russia by 2035, unless one or both of those powers expand their forces beyond their current size. Under the New START agreement of 2010, which recently expired, Washington and Moscow were limited to 1,550 such warheads. Current U.S. projections put China at about 1,500 by 2035. (Hence strategic parity, although both the Russians and Americans would likely still have larger a larger overall, factoring in non-deployed warheads.)

But another implication of these Chinese developments relates to U.S. nuclear testing. Remember that President Trump declared in October 2025 that in light of what our strategic adversaries have been doing, the United States would resume testing on an “equal basis” with Russia and China. “Because of other countries’ testing programs,” Trump said, “I have instructed the Department of War to start testing our Nuclear Weapons on an equal basis. That process will begin immediately.” 

I should stress that this “equal basis” phrasing doesn’t seem to suggest any U.S. return to full-scale nuclear weapons tests, though former Trump National Security Advisor Robert O’Brein publicly urged this back in July 2024. Nevertheless, it does suggest that we will now be willing to engage in small, yield-producing tests akin to those we have assessed that China – and indeed Russia as well, as we made clear back in 2019 and 2020 – has been doing.

For me, this is an especially interesting result, for I’ve actually been suggesting just such an approach for years – including suggesting such a thing to my State Department superiors back when we first publicly surfaced the issue of clandestine Russian tests, and our “concerns” about possible Chinese ones, six or seven years ago. More recently, I’ve even suggested this publicly, explicitly recommending in a roundtable discussion at the National Defense University in late 2024 that in the next U.S. Nuclear Posture Review, the United States 

“modify its current ‘zero-yield’ testing policy in order to permit small-scale yield-producing tests akin to those U.S. officials believe Russia and perhaps China to have conducted.” 

Why? Well, as I subsequently explained to the European Leadership Network in December 2024, 

“…[f]rom a technical perspective, there may be useful things to be learned even from extremely small yield-producing tests, but while we ourselves have foregone learning such things for decades, our adversaries apparently have not.” 

The alternative to doing such tests ourselves – as I put it in a LinkedIn post last month – is “unilaterally holding ourselves back while our nuclear adversaries advance.” Accordingly, though the issue is likely to be much debated in U.S. policy circles, I am myself supportive of President Trump’s approach on this issue:

“I wish nobody was testing at all — and perhaps we can get back to that world eventually — but I don’t want us to be sitting on our hands while our adversaries use such tests to improve their weapons. The liberty and security of too many people — in the United States as well as in many other countries — depends on the credibility of our nuclear deterrent to just give Moscow and Beijing a free pass on this.”

National security policy doesn’t have the luxury of dealing with the world as one wishes it were; we have to deal with the world we have. And in that world – the real one – it would appear that the United States is going to be getting into the business of doing small-scale yield-producing tests, at least until our adversaries convince us that they’ve stopped. I wish things were otherwise, but under the circumstances, I would say the U.S. approach makes sense.

-- Christopher Ford