US AI Data Center Supply Chain Risks from China: What Builders Need to Know

The US AI data center buildout is facing a new vulnerability: heavy reliance on Chinese-made power components. Transformers, switchgear, batteries, and optical transceivers are critical parts of the power stack, and a growing share of these parts comes from China. Washington is now moving to restrict foreign equipment, which could raise costs and slow deployment for builders.

The components at risk

Chinese firms hold significant share in several categories of data-center power equipment. According to analysts cited by CNBC, China's share in certain transformer and switchgear categories runs near 30%, and it accounts for over 40% of US battery imports. Chinese companies also lead global optical transceiver revenue, collectively supplying roughly two-thirds of units worldwide.

These components are not optional. Transformers step down high-voltage grid power to levels usable by servers and cooling. Switchgear provides centralized circuit protection. Batteries ensure uptime during grid failures. Optical transceivers carry data between facilities over fiber optic cables. The power stack is the backbone of every data center, and its dependence on Chinese imports creates a strategic vulnerability.

Policy response: executive orders and potential bans

President Trump signed an executive order declaring a national emergency over bulk-power system equipment produced abroad, authorizing the Energy Department to restrict transactions involving grid and data-center components. The FCC has also added foreign-made power inverters to its Covered List, and the administration is reportedly drafting a ban on imports of Chinese optical transceivers.

These moves mirror earlier efforts to reduce reliance on Chinese telecom gear. The Economist Group's Laveena Iyer noted that the US administration appears keen to de-risk AI infrastructure before Chinese tech is fully embedded, similar to the 5G rollout where removing Chinese equipment later caused delays and cost increases.

What this means for AI builders

For teams planning new AI data center capacity, the immediate implication is higher cost and longer lead times. Power transformers and substations are already in an estimated market shortage of 15% and 8% respectively in 2026, according to Wood Mackenzie. Restrictions on Chinese units will worsen that shortage.

An optical transceiver ban would force hyperscalers to find local alternatives. But US firms like Lumentum and Coherent, which each received $2 billion in investment from Nvidia, currently lack the capacity to scale up to Chinese volume within 12 to 24 months, per Counterpoint. This could push up costs and slow the interconnection that high-density AI workloads require.

Hitachi Energy and Siemens Energy have each announced $1 billion investments in US grid infrastructure, but these are long-term solutions. In the near term, builders should expect supply constraints and may need to pre-order components earlier or diversify sourcing. The situation is evolving, and policy details are not final. Continue monitoring but do not rely on quick fixes.

Caveats and unknowns

The analysis is based on reported dependencies and announced policy moves. Actual regulations, enforcement timelines, and China's response remain uncertain. The US data center capacity is projected to grow from 62 GW to 152 GW by 2030, but whether that growth can be sustained under component restrictions is an open question. Builders should build in flexibility and contingency plans.