Wall Street eyes utility deals as AI power race turns risky

The split in Wall Street's AI energy strategy reflects competing risk tolerances shaped by recent history. OpenAI's fossil-fuel megaproject—the most extreme expression of the Trump administration's "bring your own power" deregulation push—promises scale but carries the specter of the early 2000s energy boom-bust, when deregulation promised explosive demand that never arrived, leaving private investors with billions in losses. Jeff Jenkins of Bernhard Capital Partners lived through that cycle and is betting instead on regulated utility assets, where monopoly pricing and government oversight guarantee steady returns, if never blockbuster ones.

The market timing for this pivot is striking: utilities have spent two decades holding onto their regulated assets, but AI's capital intensity has forced major players like Duke and AEP to liquidate non-core divisions to fund data-center expansion. Private equity spotted the opening. Bernhard alone has completed half a dozen acquisitions in two years, and global PE poured $69 billion into utilities in 2025—50% above 2024—suggesting the strategy is catching on. Jenkins predicts the current building supercycle will eventually exhaust itself, after which utilities will want to buy back their old assets; he intends to sell them at a markup.

The political and environmental headwinds are real, however. Most new hyperscale projects will run on gas, raising climate concerns. And in an election year when energy prices are likely to dominate candidate messaging, regulators will face pressure to prevent AI data-center costs from being buried in ratepayer bills—a scenario that could upend the utility business model's traditional predictability.