The race to build a more self-sufficient Chinese semiconductor supply chain may matter increasingly for the computing hardware behind AI devices, but the most bullish recent forecasts describe different milestones. NVIDIA CEO Jensen Huang has suggested China could reach advanced lithography by 2030, while Elon Musk has put a broader two- to three-year window on overcoming current computing, chipmaking, and lithography constraints.
That gap in timing is less contradictory than it first appears. Huang’s 2030 comment was about advanced lithography in general and rested on China’s ability to turn a working technology into high-volume production. Musk’s assessment, made in an interview with China Media Group, addressed the wider hardware bottleneck for AI and semiconductor manufacturing rather than a specific EUV-equivalent machine.
Advanced lithography is not necessarily EUV parity
The key caveat is that neither timeline proves China will have an ASML-comparable extreme ultraviolet lithography system by 2030. “Advanced lithography” covers more than EUV, the technology used for leading-edge production and enabled by an unusually complex ecosystem of optics, light sources, masks, and manufacturing expertise.
China has made progress with immersion deep ultraviolet lithography, an older approach that can be extended toward advanced nodes through multiple patterning. SMIC has shown how far DUV can be pushed without EUV, though the method makes manufacturing more complicated, slower, and more expensive.
An EUV prototype reportedly built in China with help from former ASML engineers has generated EUV light, but it had not produced chips at the time of the report. That is a meaningful research step, yet it is far from demonstrating high-yield production of chips at scale.
A moving target for domestic chip equipment
Industry estimates illustrate the uncertainty around the finish line:
Rohmund’s view carries particular weight because Zeiss supplies the optical systems essential to ASML’s EUV machines, and he previously led the company’s EUV work. The estimates are not directly interchangeable, however: they refer to different thresholds, from improving local manufacturing capability to building an indigenous EUV platform.
ASML’s own roadmap raises the bar further. The company is considering output above 110 low-numerical-aperture EUV systems in 2028 after planning at least 80 in 2027, driven largely by AI-chip capacity expansion. It has also demonstrated a 1,000-watt EUV source, up from 600 watts, with a goal of reaching productivity near 330 wafers per hour by 2030.
U.S.-led export restrictions have blocked EUV shipments to China for years and have progressively covered some advanced DUV equipment as well. Those controls have encouraged investment in domestic alternatives across China’s chip ecosystem, including efforts tied to Huawei, SMIC, and local equipment suppliers.
China does not have to reproduce ASML’s systems component by component for those restrictions to become less consequential. Equipment capable enough for local fabs and chip designers could ease a major constraint even if it remains behind the newest Dutch tools. Whether that becomes commercially reliable production—or reaches true EUV parity—remains the material unanswered question.
Post Footer automatically generated by Add Post Footer Plugin for wordpress.
