The capital is there. The land is secured in many cases. The permits are in process. The GPU orders have been placed. Yet nearly half of the AI data center projects planned for the United States in 2026 are expected to be delayed or cancelled before the year is out. The constraint is not a chip shortage or a financing problem. It is transformers, switchgear, and batteries. The electrical equipment that connects data centers to power, distributes it safely, and stores it for reliability is in short supply and getting harder to source.
This is not a temporary disruption. The lead times for high-power transformers stretched from 24 to 30 months before 2020 to as long as five years today. AI data center deployment cycles run under 18 months. Those two numbers are structurally incompatible. Of the 12 gigawatts of US data center capacity expected to come online in 2026, only around a third is under active construction. The rest is waiting, and some of it will not wait long enough.
How the Shortage Developed
The transformer shortage did not appear overnight. The silent bottleneck in transformer and substation supply chains built over years of underinvestment in domestic manufacturing, then worsened when multiple demand sources hit simultaneously. AI data centers arrived as a massive new electrical load at precisely the moment that EV charging infrastructure, heat pump adoption, and industrial electrification were already straining the existing supply chain. All of these demand sources require the same basic components: transformers to step voltage up and down, switchgear to distribute and protect circuits, and batteries to provide backup and storage.
US manufacturing capacity did not scale to match that combined demand. The domestic transformer industry spent decades operating as a low-margin, capital-intensive business with limited incentive to build excess capacity. Now, with multi-year backlogs and order books filling years ahead, manufacturers are finally investing in expansion. Siemens Energy committed $1 billion to expand US transformer and gas turbine manufacturing. GE Vernova acquired full ownership of Prolec, a transformer company, for $5.3 billion. Those investments will ease the shortage eventually. However, they will not ease it in 2026.
The China Dependency That Nobody Wanted to Talk About
The US has spent years pursuing an onshoring agenda for semiconductor manufacturing and advanced technology components. That agenda, however, has had limited impact on the electrical equipment supply chain. China remains the world’s dominant producer of transformers, switchgear, and related components, and the US buildout depends heavily on that supply chain whether or not the companies involved prefer to acknowledge it. US imports of high-power transformers from China jumped from fewer than 1,500 units in 2022 to more than 8,000 through October 2025. Additionally, China supplies over 40 percent of US battery imports and remains close to 30 percent in some transformer and switchgear categories.
The power equipment bottleneck in the energy transition is not unique to AI infrastructure. Utilities, industrial manufacturers, and data center operators are all competing for the same constrained global supply. As a result, that competition is driving price increases, extending lead times further, and forcing developers to make procurement decisions years ahead of construction timelines that were never designed for such advance planning. A company that failed to place transformer orders two years before its construction start date is now learning that lesson at significant cost.
What Operators Are Doing to Cope
The responses to the shortage range from straightforward to genuinely creative. Amazon and Microsoft have locked in electrical equipment procurement years ahead of construction starts, built supplier relationships that provide priority access, and treated power equipment procurement as a strategic function. That approach works well for companies with the scale and capital to commit years in advance. It is less accessible for smaller operators who cannot carry the cost of pre-ordered equipment sitting in a warehouse.
The time-to-power crisis is creating a hidden scaling ceiling that capital alone cannot overcome. Some operators are refurbishing used electrical equipment to avoid new-order lead times. Others are manufacturing their own switchgear to cut supplier dependency entirely. Meanwhile, a broader wave of developers is turning to global suppliers, with Canada, Mexico, and South Korea becoming major sources of high-power transformers as domestic supply falls short. Each of these approaches carries its own cost in money, complexity, or both.
The Geopolitical Trap Inside the Supply Chain
The US data center buildout is caught in a contradiction that policymakers have not yet resolved. On one hand, the government is pushing domestic AI infrastructure investment as a national security priority. On the other, the electrical components making that infrastructure possible come predominantly from China, the same country the government identifies as its primary strategic competitor. That contradiction is not abstract. In January 2026, US utility executives touring a Chinese transformer factory noticed that roughly half the units awaiting delivery were headed to US data center projects. The dependency runs far deeper than most public discourse acknowledges.
Tariffs add cost and uncertainty without solving the underlying problem. They raise procurement costs for developers, push some toward lower-quality alternatives, and extend wait times for non-Chinese supply. Crucially, they do not create the domestic manufacturing capacity that would actually reduce dependency. That capacity takes years to build and requires sustained policy support. The current buildout provides the demand signal clearly. The policy framework to convert it into domestic supply, however, is still catching up.
Why This Matters Beyond 2026
The electrical equipment shortage will not resolve itself once the current buildout cycle concludes. The same demand drivers straining supply today, AI infrastructure, electrification, and renewable energy integration, are all structural trends that will keep growing. The 5-year wait problem is already reshaping AI infrastructure strategies across site selection, financing, and competitive positioning. Consequently, a developer that secures reliable access to electrical equipment supply chains is not just solving a logistics problem. It is building a durable competitive advantage in a market where delivering powered capacity on schedule is increasingly the primary differentiator.
US-China trade tensions are creating uncertainty around the very import channels the current buildout depends on. Tariffs and security concerns have not yet disrupted the transformer and switchgear supply chain materially, but the risk is real and growing. An industry that accepts deep dependency on Chinese electrical component imports while simultaneously expressing concern about that dependency is managing a contradiction that will eventually require resolution. Domestic manufacturing investment is the long-term answer. It will not be the short-term one, and the data centers waiting for transformers in 2026 are the clearest evidence of that gap.
