China holds a powerful position at the center of American AI data center build-out, supplying critical electrical components that American developers cannot produce in sufficient quantities.
As a result, even as U.S. tech companies pour more than $650 billion this year into AI infrastructure construction, their ambitions are limited by reliance on Chinese-made transformers, switchgear, and batteries. Although these components represent less than 10% of data center costs, they have become the determining factor in whether projects proceed on schedule, or at all.
In a striking indicator of these challenges, nearly half of the data centers slated for development in the U.S. this year are now expected to be delayed or canceled.
Extraordinary Wait Times
Trade flow numbers reveal the depth of U.S. tech dependence on China. U.S. imports of high-power transformers from China climbed from fewer than 1,500 units in 2022 to more than 8,000 units by October 2025. China also supplies more than 40% of U.S. battery imports, while its share of certain transformer and switchgear categories remains near 30%. Despite years of policy efforts aimed at reshoring manufacturing, domestic production has not kept pace with demand.
Analysts estimate that of the 12 gigawatts of data center capacity expected to come online in 2026, only about one-third is currently under construction.
Developers typically aim to bring facilities online within 18 months, yet delivery times for large transformers have expanded dramatically, from 24 to 30 months before 2020 to as long as five years today. Without these components, even completed buildings and installed computing hardware cannot operate.
To obtain rare parts, companies are scouring global supply chains, often returning to China despite geopolitical tensions and tariffs. Some developers are refurbishing older equipment or attempting to manufacture components internally, though these measures offer only partial relief.
AI Software vs. AI Hardware
While the U.S. leads in advanced chips and AI software, it has outsourced much of its electrical manufacturing base over decades. China, by contrast, controls large portions of the supply chain, from raw materials to finished equipment. That dominance makes it an indispensable supplier across the globe.
Tariffs intended to reduce dependence on Chinese imports risk slowing projects that currently rely on those imports to move forward. Developers, facing tight timelines and limited alternatives, are often willing to absorb higher costs rather than wait for domestic supply that cannot meet deadlines.
At the same time, demand for electrical equipment is rising across the enterprise and consumer markets. Grid expansion, electric vehicles, and electrified heating systems are competing for the same limited resources. This convergence is pushing prices higher and extending lead times further, reinforcing the bottleneck.
The mismatch between huge financial investment and limited physical infrastructure is raising concerns that planned AI capacity will not materialize at the anticipated pace. In sum, without a significant expansion of domestic manufacturing or a stable policy for continued imports, the American AI build-out remains tied to external supply chains it cannot control.
