U.S.-China AI supply chains emerge as next battleground

U.S.-China AI supply chains emerge as next battleground
AI supply chains heat up

A fragile truce in U.S.-China trade and technology tensions is nearing its expiry as competition over artificial intelligence shifts from headline chip bans to the supply chains behind them. The change is raising risks for data centres, chip equipment, rare earths and server components that underpin AI growth across the global technology sector.

Highlights

  • Expiration of the Busan truce in January and no extension at the Shenzhen summit shift U.S.-China AI competition focus to supply chains, with critical minerals and hardware as key pressure points.
  • New U.S. licensing requirements for offshore data centres providing AI capabilities to Chinese firms could impact ByteDance, Alibaba, Tencent, and Moonshot if enforced more strictly.
  • Anthropic’s $1.5 trillion public listing highlights widening valuation and supply advantages for U.S. AI firms over Chinese peers, as global electronics supply chain interdependence increases market vulnerability.

Truce strains expose supply chain vulnerabilities

As reported by Financial Times, the Busan truce that has steadied relations since October 2025 is due to expire in January, with Beijing and Washington increasingly testing each other in areas tied to AI hardware, critical minerals and overseas computing capacity. This week’s summit in Shenzhen ends without an extension, reinforcing expectations that the next phase of rivalry centers on supply chains rather than only tariffs or model development.

Both governments have used the pause to seek strategic leverage. The U.S. has tried to reduce reliance on Chinese rare earths while preserving its lead in advanced chips and chipmaking tools, and China has sought to cut chip dependence and expand influence through open-weight AI models. Yet the truce never fully removed pressure points, as China tightened restrictions affecting materials such as yttrium and the U.S. kept restrictions and licensing measures in place across selected technology categories.

Washington still appears to hold powerful choke points through export controls on advanced lithography and other semiconductor equipment. China remains dependent on foreign suppliers, especially ASML, for the most advanced tools needed to produce AI chips, while tighter limits on SMIC have constrained processor output even as memory-chip producer CXMT expands capacity more quickly.

That imbalance has fed reliance by Chinese AI groups on offshore data centres filled with Nvidia chips, as well as illicit channels, to secure computing power. New U.S. licensing requirements for data centres outside China that provide AI capabilities to Chinese companies could therefore hit major users including ByteDance, Alibaba, Tencent and Moonshot if enforcement tightens further.

Global technology and trade face wider fallout

The broader industry impact extends far beyond the two countries because AI manufacturing depends on a dense cross-border network of minerals, gases, boards, optical links and servers. With 1.3 million components in a single Nvidia Vera Rubin server, any disruption could spread unpredictably through production lines, pricing and delivery schedules across Asia, North America and Europe.

Japan has already warned that Chinese restrictions are pushing some companies toward shortages of yttrium used in chipmaking tools and multilayer ceramic capacitors for AI servers. At the same time, the U.S. is promoting alternative supply routes through initiatives including Pax Silica with partners such as Japan and South Korea, though replacing entrenched capacity will require years of investment and execution.

Financially, the divergence between U.S. and Chinese AI ecosystems is becoming more visible. Anthropic’s public listing at a $1.5 trillion valuation contrasts sharply with lower valuations and weaker revenues among Chinese peers, reflecting how hardware access, capital markets and international restrictions are shaping competitive outcomes as much as software innovation.

Even so, the two economies remain deeply intertwined through Taiwan and broader east Asian electronics networks. The U.S. is importing rising volumes of AI server-related goods from Taiwan, while China continues supplying components used in many of the same systems, leaving global growth increasingly exposed to the very supply chains now turning into the central arena of U.S.-China competition.

We recently wrote about older chip exports as Washington signaled only older-generation semiconductors should be sold to China.

This material may contain third-party opinions, none of the data and information on this webpage constitutes investment advice according to our Disclaimer. While we adhere to strict Editorial Integrity, this post may contain references to products from our partners.