After spending more than $150 billion to achieve semiconductor self-sufficiency, China is still producing far fewer and weaker chips than global rivals, exposing a major bottleneck in its AI ambitions.

At a recent Beijing conference, executives said China’s AI sector could soon rival global leaders. Yet one obstacle remains: access to high-performance chips comparable to those from Nvidia.

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

Why China Is Still Behind

  • Chinese manufacturers will likely produce only about 2% as many advanced AI chips as foreign competitors this year.
  • The biggest constraint is lack of access to critical equipment from ASML, whose machines are essential for cutting-edge chipmaking.
  • Export controls backed by Washington have blocked Chinese firms from buying that technology.

China’s top foundry, SMIC, can produce chips domestically, but analysts say yields are lower, defect rates are higher, and power efficiency trails industry leaders like TSMC.

Huawei’s Central Role

National champion Huawei has led the domestic push, designing processors for smartphones and AI systems. Its latest chips rival some older Nvidia models, but experts say many still rely on foreign components or manufacturing steps.

To compensate for weaker chips, Chinese firms are linking large numbers of them together in massive computing clusters. The approach works but is expensive and energy intensive.

Policy Pressure Accelerated the Push

China’s self-reliance drive intensified after US restrictions and sanctions during the first administration of Donald Trump cut off companies from key technology. Those actions convinced Beijing it needed its own full semiconductor supply chain by 2030.

Ironically, Trump recently allowed limited chip sales to China again, offering temporary relief while negotiations continue.

China’s strategy is producing progress but not parity. The country has built a vast domestic chip ecosystem, yet it still depends on foreign technology for the fastest processors. Until it closes that gap, its AI ambitions remain constrained by hardware limits.

Disclosure: This article does not represent investment advice. The content and materials featured on this page are for educational purposes only.