Trillions in planned spending on data centers are running up against chronic shortages of transformers, switchgear, and batteries, the unglamorous gear that actually delivers power to the racks. 

Domestic production has not scaled anywhere near fast enough, leaving developers with little choice but to lean on overseas suppliers, predominantly from China. The result is lengthening lead times that threaten to push back or cancel projects already baked into corporate budgets and national strategy.

Bloomberg reports electrical equipment, though a small slice of total project costs, is the component that can bring everything to a halt. Their leading example is the massive facility under construction in Abilene, TX, expected to draw as much as 1.2 gigawatts once it serves OpenAI.

For comparison, that's more energy than a Westinghouse AP1000 reactor can provide... 

We previously pointed out exactly this vulnerability back in August 2025 when Wood Mackenzie sounded the alarm on transformer shortages. The consultancy projected demand would exceed supply by 30 percent that year alone, with U.S. manufacturers able to cover only a fraction of needs and roughly 80 percent of units imported. We warned then that the AI boom was colliding with a grid already buckling under failed green policies and surging electricity loads, a dynamic that has only intensified since.

In January, we highlighted America's aging power infrastructure, showing how data-center demand is now a measurable slice of national consumption and exposing decades of underinvestment that no amount of policy rhetoric can paper over. 

The current administration is doing what they can to try and ensure costs are not passed on to household consumers, with the recent agreement made between the White House and some of the biggest hyperscalers. But with news from Constellation that no matter how hard new energy generation is pushed onto the grid, long connection queues will destroy any possibility of the national grid finding balance.