Half of planned U.S. data center construction has been delayed or canceled, with growth limited by a lack of power infrastructure and parts from China — the AI ​​build-out flips the breakers.

When you make a purchase through links in our articles, Future and its syndication partners may earn a commission.

The trade war between the U.S. and China has forced server makers to pull out of the People’s Republic, reducing U.S. companies’ reliance on Tianxia manufacturers. However, China has become the world’s largest producer of the electrical equipment needed to build the electrical infrastructure inside and outside AI data centers. To that end, a lack of power delivery equipment, including devices from China and other countries, is slowing the project’s timeline, Bloomberg reports.

Despite unprecedented investment in AI infrastructure — Alphabet, Amazon, Meta, and Microsoft are expected to spend more than $650 billion in 2026 to expand AI capabilities — half of U.S. data center construction is projected to be delayed or canceled this year. Bloomberg. A major reason behind these constraints is the availability of key electrical components – such as transformers, switchgear and batteries – that are used both on and off data center sites, as AI companies must extend the grid infrastructure to supply sufficient power to their data centers. Meanwhile, grid infrastructure is also being emphasized by electric vehicles and electrified heating systems.

About 12 gigawatts (12 GW) of data center capacity is expected to come online in the U.S. by 2026, according to data cited by market intelligence firm Sightline Climate. Bloomberg. However, only one-third of that capacity is currently under active construction due to various constraints.

Electrical infrastructure represents less than 10% of total data center costs, but it is just as important as compute hardware. A delay in any single element of the power chain can bring the entire project to a halt, making transformers, switchgear and similar devices critical items despite a relatively small portion of CapEx.

Due to high demand, lead times for high-power transformers have expanded dramatically in the U.S.: Delivery typically took 24 to 30 months before 2020, but wait times can be as long as five years today, as cited by Sightline Climate. Bloomberg. For AI data centers, this is a disaster because their deployment cycle is less than 18 months.

To address the shortage, companies are turning to global markets. As a result, Canada, Mexico and South Korea became the largest suppliers of high-power transformers for AI data centers in AI data centers. At the same time, imports of high-power transformers from China rose from less than 1,500 units in 2022 to more than 8,000 units in 2025 through October, according to data cited by Wood Mackenzie. Bloomberg.