Regulation

China Starts Mass-Producing Immersion DUV Lithography Tools

The Information reports China now mass-produces immersion DUV lithography machines, delivering to SMIC, Hua Hong, and CXMT this year — 5 tools in 2026, 20 in 2027; ASML fell 8%.

China Starts Mass-Producing Immersion DUV Lithography Tools — article cover

On July 27, The Information reported, citing two people familiar with the program, that a state-backed Shanghai company has begun mass-producing immersion deep ultraviolet (DUV) lithography machines. The market reacted the same day: ASML shares fell as much as 8%. First deliveries are expected this year to China’s biggest chipmakers — SMIC, Hua Hong Semiconductor, and Changxin Memory Technologies (CXMT) — with targets of roughly 5 machines in 2026 and about 20 in 2027.

From Assembled Teams to a Production Line

The Information did not name the manufacturer. Its sources said the operation pulled DUV development teams from several Chinese firms, including Shanghai Yuliangsheng Technology; a separate Tom’s Hardware report attributes China’s first domestic immersion DUV scanners to Shanghai Aishengna. SMIC has been testing a Yuliangsheng immersion tool since September 2025.

The customer list is itself the message. SMIC, Hua Hong, and CXMT are exactly the entities a US House resolution (H.R. 8170) would designate as restricted. Domestic tool production means the supply line that export controls pinched now has a substitute.

Single Exposure Prints 28nm; 7nm Takes Multipatterning

Immersion DUV prints 28nm-class features in a single exposure. Getting to 7nm requires multipatterning, at a cost in overlay errors and yield — every extra patterning pass adds steps, and every added step eats margin. Most components in the Chinese tools are domestic, but some critical parts still come from Japan, and delays from local suppliers have slowed the 2026 output target.

The comparison point is ASML: roughly 130 immersion systems shipping in 2026, a 98.7% share of the immersion market, plus a plan to expand immersion capacity by another 30% in 2027 and a further 30% under evaluation for 2028. Five versus one hundred thirty — the gap has to be read in multiples, not percentages.

From “Can Build” to “Production-Ready”

Nick Patience, AI lead at Futurum Group, put it directly: Chinese chipmakers need “to get to at least yield parity, not just have a working tool.” He noted that yields from SMIC’s imported DUV machines are still far behind TSMC’s, and that reliability takes years of iteration in the field. SemiAnalysis called scaling production “the most underestimated part,” pointing to tool performance, fleet performance, ecosystem gaps, and poor economics against ASML machines that are already fully depreciated.

Analysts advising caution go further. Stephane Houri of ODDO BHF said to take the news “with a pinch of salt,” since what China produces “could be limited to the very low end.” Paul Triolo of DGA Albright Stonebridge Group said matching ASML would require producing “reliable machines serially” with global fab support — “quite a stretch.” The AI Futures Project’s June assessment was more conservative still: commercial-scale Chinese immersion DUV lands in the mid-2030s.

What It Means for Controls and the Supply Chain

For ASML, the short-term hit is limited. In SemiAnalysis’ words, the Chinese tools “displace revenue ASML already lost to export controls” — “a tool ASML cannot legally or physically supply being built locally does not subtract from a sold-out order book.” The stock is still up more than 123% this year. China accounted for 14% of ASML’s Q2 net system sales of €6.6 billion, and about 20% of sales for the year, down from 33% in 2025.

The action is on the policy side. Beyond H.R. 8170, the MATCH Act — introduced in April, reported out of the House Foreign Affairs Committee on April 22, with a Senate companion in S.4281 — would extend controls to servicing and technical assistance on already installed tools. As for EUV: Reuters has reported a working Chinese prototype, but ASML took roughly two decades and about $10 billion in R&D, co-funded by Intel, TSMC, and Samsung, before EUV turned profitable at scale around 2018-2019. Triolo’s conclusion: “I think EUV is out of reach. Never say never… but it’s a completely different technology.”

The number to watch is not 2026’s five machines but whether the 2027 target of twenty holds. That is when serial production, not prototypes, gets tested — and when the yield question moves from analysts’ caveats to fabs’ own data.

Sources

AI-assisted summary compiled from the sources above, reviewed by a human before publishing.

FOUND_THIS_USEFUL?

Support more practical AI articles, tutorials, and build notes.

BUY_ME_A_COFFEE
SHAREXEMAIL