A look at the evolution of China's chip industry and Huawei under US export controls, as GPUs have become the new focal point of US-China tech competition
Paul Triolo / American Affairs Journal :
Context & Ripple Effects
Earlier coverage characterized the initial US controls as a severe constraint on China’s semiconductor and AI ambitions. By 2024, Huawei had responded by expanding into new businesses and building out its supply chain with Beijing support, as described in Huawei’s supply-chain buildout.
This analysis matters because it identifies GPUs—not chips broadly—as the central contestable input, connecting Huawei’s adaptation to the broader use of compute chokepoints in US-China technology policy.
First-order effects
- US export controls directly limit Huawei’s and China’s access to leading GPU and chipmaking inputs, making domestic alternatives and supply-chain resilience more immediately important.
- Huawei becomes a more consequential focal point for China’s response: its ability to substitute for restricted technology affects both its own operations and the wider domestic chip ecosystem.
Second-order effects
- Restrictions on Nvidia, AMD, and Intel sales to China can create additional room for Huawei’s offerings, a dynamic later highlighted in coverage of tighter AI-chip export restrictions.
- Chinese chip suppliers and AI developers face stronger incentives to align around locally available compute, while US chip vendors must navigate a market shaped by licensing and product-specific controls.
Third-order effects
- The competition shifts from access to individual chips toward control of the compute stack: accelerators, manufacturing capabilities, supply chains, and the systems built around them.
- If controls continue to steer demand toward domestic substitutes, export policy may simultaneously constrain China’s frontier access and accelerate the commercial relevance of protected local suppliers; the balance depends on their ability to close technical gaps.
The trend: GPUs are becoming strategic leverage in a broader shift from semiconductor trade toward contested national compute ecosystems.