China entering mass production of indigenous chip manufacturing equipment, reducing ASML dependence.
China has begun mass-producing its own deep ultraviolet lithography machines, and ASML investors just watched roughly $44 billion in market value disappear over it.
On July 27, reports surfaced that a Shanghai-based consortium had started shipping domestically built immersion DUV scanners to Chinese chipmakers. It marks a milestone Beijing has pursued since Washington and the Netherlands tightened export controls on advanced lithography equipment. ASML shares fell as much as 8% that day, according to Bloomberg—the steepest single-session decline for the Dutch equipment maker since early July.
The tools reportedly come from Shanghai Aishengna Electronic Technology Group, assembled by combining engineering teams from Shanghai Yuliangsheng Technology, Huawei-linked SiCarrier, and personnel and technology transferred from SMEE, China's national lithography champion. SMIC began testing early units in September 2025. Both CNBC and Asia Times reported this week that the machines are transitioning from prototype to limited production runs destined for SMIC, Hua Hong Semiconductor, and memory maker ChangXin Memory Technologies.
This is no EUV breakthrough. These are immersion DUV, ArFi machines rated for 28-nanometer processes—the mature nodes found in automotive and industrial equipment, not the sub-3-nanometer nodes powering Nvidia's latest GPUs or Apple's newest phones. China claims the tools could reach 7-nanometer through multi-patterning, a technique that stacks multiple exposure passes to achieve finer detail. Asia Times reported the domestic machines remain roughly four generations behind ASML's current DUV lineup. EUV, which ASML effectively monopolizes and which China needs for top-end AI chips, is an entirely different technology.
Scale heavily favors ASML. Chinese production plans target approximately 5 units this year and roughly 20 in 2027, according to eWeek. ASML alone expects to ship about 130 DUV immersion systems in 2026, with a planned 30% increase for 2027. The disparity underscores what "mass production" actually means here: limited volume.
Not everyone treats the milestone as transformative. SemiAnalysis noted the real test isn't lab performance but whether tools can sustain "tool performance, scaling production, fleet performance, surrounding ecosystem and poor economics against fully depreciated ASML machines"—equipment Chinese fabs have already purchased and operated cheaply for years. Dan Hutcheson, vice chair at TechInsights, was blunt: "The harder they run, they just stay in place." A Futurum Group analyst highlighted what truly matters to foundries: yield. Running a machine once is one achievement. Achieving yield parity with ASML's tools run after run is another.
ASML's China revenue share has declined from 33% to roughly 20% of 2026 total sales as US and Dutch export limits tighten. The company's CEO has warned that further controls would only accelerate Beijing's competing tool development, weakening rather than protecting Western firms. That forecast now reads less hypothetical and more prescient. Bank of America called the selloff an overreaction, arguing ASML carries more than 70% upside potential and faces limited near-term financial risk from a 28-nanometer tool designed for mature nodes. ASML is guiding to $49 billion to $51 billion in 2026 revenue, and its CFO has noted the company maintains rare-earth stockpiles accumulated over years of extended lead times—a buffer against supply shocks of this nature.
If you've been told export controls would keep China a decade behind, here's the exception to monitor. Sanctions delay capability; they don't eliminate it. China cannot yet access EUV, and the frontier AI chips dependent on it remain beyond reach. But a domestic DUV supply chain now feeding SMIC, Hua Hong, and CXMT means China no longer requires ASML's authorization to sustain its mature-node fabs. That is the leverage US policy was designed to deny Beijing—and it is slipping away, one 28-nanometer wafer at a time.