When the Grid Can’t Keep Up

Did you know the U.S. is facing a significant shortage of electrical power? Demand from AI hyperscalers and data centers has surged, and the Department of Energy warns that if new generation doesn’t keep pace with plant retirements, blackout risk could rise dramatically - with some scenarios showing up to a 100-fold increase by 2030. AI and data-center loads currently account for roughly 4–5% of U.S. electricity use, and several forecasts project them rising toward 8–10% by 2030. In its current state, the U.S. power grid cannot fully meet present or future demand.

Government agencies and AI industry leaders are increasingly concerned that the U.S. may lose its competitive edge to China, which now has well over twice the U.S. generation capacity and is expanding it at a far faster pace. This widening gap could influence which countries are best positioned to support energy-intensive AI growth.

We’re already seeing the effects of an oversubscribed grid in commercial real estate. In some regions, utilities have reallocated or downgradedelectrical service to buildings that sat empty during the pandemic, prioritizing areas with active demand. As a result, historical power specs often do not match what the utility can deliver today, especially as AI demand accelerates faster than new generation can be built under current federal energy policy. This creates real risk for occupiers signing leases under outdated assumptions.

A good tenant advisor will confirm with the utility what capacity is actually supported today, rather than relying on outdated building specs. It’s not about inspecting transformers ourselves - it’s about making sure the power a lease promises still exists. This issue isn’t going away, and CRE decision-makers will need to adapt as these constraints tighten.