China did not break ASML's monopoly. Washington may be building the customer that one day does
Five Chinese lithography machines knocked roughly 55 billion euros off ASML in two sessions. The headlines pointed at the wrong technology, and the fear underneath them is justified anyway.
On 27 July 2026 The Information reported, and Reuters confirmed the next day, that a state-backed Shanghai manufacturer had started producing immersion DUV lithography machines: about five this year, roughly twenty in 2027, destined for SMIC, Hua Hong and the memory maker CXMT. ASML fell between about 6% and 8%, a swing of some 55 billion euros in market value, roughly six times what the company expects to sell in China all year.
Twelve days earlier it had looked like the opposite story. On 15 July ASML reported 9.3 billion euros of quarterly sales, raised full-year guidance to between 43 and 45 billion, and said it was adding about 30% more DUV immersion capacity in 2027 with another 30% under study for 2028. A company that cannot build fast enough became, in under a fortnight, a company whose monopoly was declared over.
The headlines were pointing at the wrong machine, and the correction is narrower than it looks. ASML's true monopoly is EUV, and no EUV system has ever crossed into China. What China built is immersion DUV, a generation older, still indispensable, and precisely the range ASML was still allowed to sell there. The market's fear was not irrational. It was just aimed at the wrong quarter.
What China actually built, what it still cannot do, and where the real risk sits
1. The company is three years old and is a merger of everything China already had
Shanghai Aishengna Electronic Technology Group was registered in August 2023 with 7 billion yuan of capital, roughly a billion dollars, and two state shareholders: Shanghai Electric Holding and a subsidiary of Shanghai International Trust. It absorbed teams from Yuliangsheng, the lithography startup linked to the Huawei-backed toolmaker SiCarrier, which SMIC has been testing an immersion prototype from since September 2025, and from SMEE, the veteran of Chinese lithography. Corporate and recruitment records place Aishengna and Yuliangsheng at the same Shanghai address. This is not a handful of engineers who suddenly invented a scanner. It is a national team assembled under sanction pressure, which is a harder thing to compete with.
2. Five machines, measured against the right denominator
ASML shipped 131 immersion systems in 2025 and is building to a capacity of around 130 for 2026, with 30% more planned for 2027. Five units is under 4% of that, and twenty in 2027 is roughly 12%. Analysis published after the selloff put twenty domestic tools at about 1.4 billion euros of displaced ASML sales, some 2.4% of group revenue. The machines are described in the reporting as not yet production-qualified, needing months of further testing, mostly built from domestic components with some critical parts still coming from Japan.
3. The specification that matters is not resolution, it is repeatability
The machine reportedly targets 28 nanometres in a single exposure, with 7 or even 5 nanometres reachable through multipatterning, passing over the same layers repeatedly to get finer than one pass allows. Slower, costlier, worse yields, and exactly how SMIC already makes 7 nanometre chips for Huawei without any EUV. The comparison usually skipped: the SMEE SSA800 platform these teams come from runs around 150 wafers an hour, while ASML's TWINSCAN NXT:2150i is specified at 310 or more, with 1.0 nanometre matched machine overlay. Overlay is the alignment error between the hundred-odd stacked layers of a modern chip; matched means any two tools in the fleet can be swapped without it drifting. A fab does not buy a machine. It buys a fleet that behaves identically for years.
4. Nobody can compare the fleets, because there is not a Chinese fleet yet
There is no published throughput under production conditions, no availability figure, no yield, no maintenance record, because no Chinese immersion fleet has yet run for months across several fabs. That is a stage, not an insult, and the stage is long. ASML shipped its first EUV alpha demo tools to IMEC and SUNY Albany in 2006, delivered an NXE:3100 to TSMC in 2010, and the first commercially available EUV process, TSMC's N7+, only reached volume production in 2019. Thirteen years, with ASML, TSMC, Samsung and Intel all pulling, billions spent, and not one sanction in the way.
5. The reassuring version of this story has a hole in it
The tidy summary doing the rounds is that EUV is the monopoly and DUV never was, because Nikon sells DUV too. On dry DUV that is fair. On immersion it is not: ASML is estimated to hold about 98.7% of the immersion market, and shipped 131 of those systems last year. So China did not walk into a competitive segment. It stepped into the one part of the near-monopoly that export controls had left it any reason to attack.
6. Export controls built the business case, and the researchers say so plainly
Before the restrictions no rational fab manager would buy a slower, less precise, unproven domestic tool while ASML would still sell them one. The Netherlands introduced licensing for advanced DUV from 1 September 2023 and extended it on 7 September 2024 to the NXT:1970i and 1980i, under years of American pressure and after Washington had begun applying its own rules to Dutch machines containing US components. Sanne van der Lugt of the Leiden Asia Centre put the result bluntly: "What this shows is that U.S. export controls successfully created a business case for Chinese lithography. It was not what they were intended to do, but that is the outcome." And, on the customer's side of it: "For Chinese customers it may be this or nothing."
7. The real exposure is not a better machine. It is losing the right to service the installed ones
On 22 April 2026 the House Foreign Affairs Committee reported H.R. 8170, the MATCH Act, by 36 votes to 8. It is not law: the House has not passed it on the floor and the Senate companion, S. 4281, remains in committee. What it would do is designate every fab operated by SMIC, CXMT, YMTC, Hua Hong and Huawei as a covered facility, apply a presumption of denial to exports and to servicing, and give allied suppliers, the Netherlands and Japan among them, 150 days to match, failing which US jurisdiction extends over allied tools built with American technology. Servicing is the word to watch, because it covers installation, calibration, repair, refurbishment, diagnostics, firmware and software updates, training, technical assistance and process tuning. Nearly everything that keeps a complex machine at specification for a decade.
8. That is what flips the economics inside a Chinese fab
A fully depreciated ASML scanner already on the floor, with staff trained on it and processes calibrated to it, is the hardest competitor a domestic tool can face, not because it is free to run but because the capital was spent years ago. The Chinese machine never needed to be as good as an ASML. It only needs to become the best machine the customer is confident of still having next year. Erode the support for the installed base and the incumbent's advantage decays quietly, without a single new machine being sold to anyone.
9. Beijing is not stopping at DUV, and scepticism has been wrong before
Reuters reported in December 2025 that a high-security laboratory in Shenzhen, staffed in part by former ASML engineers working under false identities and coordinated through a Huawei-led programme, had completed an EUV prototype built from salvaged and secondary-market parts. It fills nearly a factory floor, it generates extreme ultraviolet light, which is hard enough on its own, and it has produced no working chips. The official target for chips is 2028; people close to the project suggest 2030. Reflexive dismissal of Chinese capability has a poor record here: SMIC's 7 nanometre process for Huawei was also supposed to be out of reach without EUV.
10. Beijing read the share price correctly
The Global Times, part of the People's Daily group, wrote that the fall in chip equipment shares showed "Western investors' blind faith in the effectiveness of blockades has been shaken, and the US-led decoupling strategy has lost much of its momentum." That is state media doing its job, and it is also a fair description of what happened. Nobody repriced ASML by 55 billion euros over five machines worth a fraction of a percent of its output. They repriced the moment substitution stopped being hypothetical.
So both halves are true, and almost everyone is picking one. ASML has not lost its EUV monopoly: no other company on earth sells a machine that prints the finest layers of an advanced chip at industrial scale, and nothing announced this summer changes that. China has taken a real step toward substituting Western tools inside its own market, on the one technology it was permitted to need. The West spent years stopping China from buying the right machines. The more interesting possibility is that it has been building the captive market a domestic industry needed in order to exist at all.
The American bet is about time rather than permanence. Brian Mast, who chairs the committee that moved the MATCH Act, argues that without these tools China cannot make its own advanced chips and will not be able to for years, which may well be right; slowing an opponent is a more plausible goal than stopping one with the resources to keep going. That is the trade being made: accept a medium-term competitor to protect a short-term lead, and wait a decade to find out whether it was worth it. For everyone downstream, the practical signal is simpler. Nothing here changes what a chip costs this year, and the day the Shenzhen machine or its descendant prints wafers at industrial throughput and yield is the day the question is worth reopening. Until then, watch how hard it becomes to keep the ASML machines in China running, not how good the Chinese ones are.
Questions people ask
- Did China break ASML's monopoly?
- No. ASML remains the only company selling EUV lithography systems at industrial scale, and no EUV system has ever been shipped to China. The reports of July 2026 concern immersion DUV, an older generation of machine that is still essential to most chipmaking. China producing its own immersion DUV tools is a genuine industrial achievement and it does not touch the EUV monopoly.
- What is the difference between EUV and DUV lithography?
- Both print circuit patterns onto silicon wafers, using different light. DUV, deep ultraviolet, uses 193 nanometre light, and immersion DUV puts a layer of water between the lens and the wafer to resolve finer features. EUV, extreme ultraviolet, uses 13.5 nanometre light and is required for the finest layers of the most advanced chips. ASML is the only industrial supplier of EUV; on DUV it competes with Nikon, though it holds an estimated 98.7% of the immersion segment.
- How much of ASML's revenue comes from China?
- The company expects China to account for about 20% of revenue in 2026, roughly 9 billion euros, down from around 33% in 2025. The decline reflects Dutch export licensing introduced in September 2023 and extended in September 2024, which restricts sales of the most advanced immersion systems to Chinese customers.
- Can China make 7nm chips without EUV?
- Yes, and it already does. SMIC produces 7 nanometre chips for Huawei using DUV immersion tools and multipatterning, which passes over the same layers several times to reach features finer than a single exposure allows. The penalty is real: more steps, more time, higher cost and lower yields, which limits how much can be produced economically.
- What is the MATCH Act?
- H.R. 8170 and its Senate companion S. 4281, the Multilateral Alignment of Technology Controls on Hardware Act. It would apply a presumption of denial to exports and servicing of chipmaking equipment for fabs operated by SMIC, CXMT, YMTC, Hua Hong and Huawei, and give allied suppliers such as the Netherlands and Japan 150 days to match US restrictions. The House Foreign Affairs Committee reported it 36 to 8 on 22 April 2026. It is not law: the House has not passed it and the Senate bill is still in committee.
Written from an owner-supplied French-language video essay on China's lithography programme, and checked on 13 September 2026 against Reuters's report of 28 July 2026 and its syndications, Bloomberg and CNBC coverage of the 27 July selloff, ASML's own second-quarter results of 15 July 2026 and its published TWINSCAN NXT:2150i specifications, Tom's Hardware, TrendForce and Implicator summaries of The Information's original report, the Global Times editorial of 28 July 2026, Reuters's December 2025 report on the Shenzhen EUV prototype, GovTrack records for H.R. 8170 and S. 4281, committee and law-firm summaries of the MATCH Act markup of 22 April 2026, TrendForce and ASML statements on the Dutch export licensing of September 2023 and September 2024, and ASML's own account of the EUV timeline from the 2006 alpha demo tools to TSMC's N7+ reaching volume production in 2019. Corrections to the source material: the market-value loss is reported at around 55 billion euros rather than more than 60 billion, with roughly $44 billion wiped from the US-listed shares in the session; the claim that the targeted segment "was never really a monopoly" holds for dry DUV, where Nikon competes, but not for immersion, where ASML is estimated at about 98.7% of the market; the NXT:2150i matched machine overlay is specified at 1.0 nanometre rather than under a nanometre; the Shenzhen EUV prototype was completed in early 2025 and reported by Reuters in December 2025, not built in December; and ASML's EUV road to volume production ran from 2006 to 2019, thirteen years rather than "more than a decade" ending in 2018, which was the year of risk production for TSMC's N7+. Two company names appear in the reporting because two outlets named different entities: The Information pointed to Shanghai Yuliangsheng, Reuters to Shanghai Aishengna, which absorbed Yuliangsheng's teams; the programme is deliberately opaque and neither account has been confirmed by the companies. Labelled as reported rather than verified: the five and twenty machine figures, the customer list of SMIC, Hua Hong and CXMT, the 2028 and 2030 EUV targets, and the estimate that twenty domestic tools would displace about 1.4 billion euros of ASML sales. Not verified independently: the source's statement that Aishengna has no identifiable website. TaskNorth holds no position in any company named here.
Trying to work out which AI tools fit your task? Describe the outcome and get a Blueprint: the tools, the prompt, and the steps, with pricing we verified ourselves.