How China built its ‘Manhattan Project’ to rival the West in AI chips

Original Article ↗ Hacker News Discussion ↗

Technical achievement vs. production reality

  • Many commenters stress the lab has “a working EUV light source,” not a full, production EUV scanner.
  • The really hard parts are said to be mirrors/optics (Zeiss-level), masks, wafer positioning at nanometer accuracy and high throughput, and long‑term reliability in fabs.
  • Question raised: how far is “generates EUV light” from “production‑ready tool”? Consensus: still far; 2028–2030 for usable chips is seen as plausible but not guaranteed.
  • China is already competitive economically at 7/5 nm via DUV multi-patterning and cheap energy; EUV is about catching up on efficiency and future nodes.

ASML’s moat, export controls, and ecosystem

  • Strong view that ASML’s true moat is its ecosystem: Zeiss optics, Cymer light sources, global suppliers, and decades of integration/yield tuning.
  • Debate over how much leverage the US has via export controls on Cymer and the EUV light source, and whether ASML could “recreate” that capability in Europe.
  • Some argue the US–Netherlands setup is an intentional, deeply entangled security partnership; others imagine scenarios where geopolitical rupture would break US control.
  • Smuggling of legacy DUV tools is discussed and mostly dismissed as limited and conspicuous; EUV tools are seen as nearly impossible to move covertly.

Talent, “reverse engineering,” and security

  • Project is widely believed to rely on former ASML engineers (often Chinese-born) recruited with large bonuses and secrecy measures (fake IDs, aliases).
  • Disagreement over whether this is normal labor mobility vs. de facto industrial espionage.
  • Some call for sanctions; others note these engineers have already accepted that their careers are now China‑bound.
  • Broader concern about Chinese nationals (and “true believers” of any origin) in sensitive Western orgs vs. the risks of ethnic profiling and discrimination.

Economic and hardware implications

  • Expectation that once China has “good enough” domestic EUV/DUV, it can undercut Western suppliers by treating advanced lithography as a low‑margin utility.
  • That could compress Western semiconductor margins and force more subsidies or R&D cuts.
  • Many hope Chinese GPUs and AI chips will counter Nvidia’s data‑center focus and bring cheaper consumer hardware; others worry about trust and opaque firmware on China‑sourced silicon.

Geopolitics, Taiwan, and strategy

  • Some see this as reducing China’s dependence on TSMC and thus lowering Taiwan’s deterrent value; others say Taiwan’s status is driven more by ideology and legitimacy than chips.
  • Competing scenarios: military invasion this decade vs. “buying” or economically absorbing Taiwan by flooding the world with cheap high‑end chips.

Framing: “Manhattan Project” and West–China narratives

  • Divided views on the title: some find the nuke analogy sensationalist, especially in a Japanese outlet; others say “Manhattan Project” is now just shorthand for a massive, state‑backed R&D push.
  • Several commenters argue Western media and commenters still underestimate China’s speed in catching up once a target is set, drawing analogies to the Soviet bomb and to China’s rise in EVs, solar, and other industries.