Here's a twist in the AI investment story that nobody saw coming from a political podium. In his recent State of the Union address, President Donald Trump didn't just talk about artificial intelligence; he talked about its power bill. And he suggested a radical shift: the companies building the giant data centers that fuel AI should be the ones to pay for—and possibly generate—their own electricity.

"We have an old grid, it could never handle the kind of numbers—the amount of electricity—that's needed," Trump said. "So I'm telling [companies] they can build their own plant; they're going to produce their own electricity." He paired this with a new "ratepayer protection pledge," claiming that "no prices will go up, and in many cases, energy prices will go down for communities."

For ETF investors who had been placing bets on the straightforward premise that AI equals more electricity sales, this introduces a fascinating new variable. The game might be changing.

Utilities ETFs: A Detour on the AI Highway?

The thesis was simple and compelling. Electricity demand from data centers has doubled since 2018 and, by federal estimates, could triple again by 2028. After years of stagnant growth, this AI-driven surge was a dream scenario for utility companies. ETFs like the Utilities Select Sector SPDR Fund (XLU), the Vanguard Utilities ETF (VPU), and the iShares U.S. Utilities ETF (IDU) had ridden this wave, growing between 16% and 19% over the past year and around 9% each year-to-date.

But what if the biggest customers decide to generate their own power? Hyperscalers—the tech behemoths like Alphabet Inc. (GOOGL), Amazon.com, Inc. (AMZN), and Meta Platforms, Inc. (META)—are already exploring captive power solutions. These can range from natural gas plants and long-term nuclear contracts to sophisticated renewable energy setups paired with massive battery storage. If they go all-in on self-generation, a significant chunk of that anticipated demand growth might simply bypass the traditional, regulated utility.

The risk here is clear: investors may have overestimated how directly AI translates into higher electricity sales for the companies in those popular utilities ETFs. The growth engine might sputter if the fuel is coming from somewhere else.

The Infrastructure Angle: Follow the Wires

But here's where it gets interesting. The problem—and the opportunity—might not be about the electrons themselves, but about the wires that carry them.

Think about it. Even if Amazon builds its own natural gas plant or Meta erects a solar farm, that power still needs to get to the data center. It needs transmission lines, substations, transformers, and grid interconnections. This infrastructure is often subject to regulated structures and becomes part of a utility's "rate base"—the assets on which they are allowed to earn a return. Building this stuff is a capital-intensive business, and someone has to pay for it.

This dynamic could quietly benefit ETFs focused on the nuts and bolts of the grid. Funds like the First Trust NASDAQ Clean Edge Smart Grid Infrastructure Index Fund (GRID) and the Global X U.S. Infrastructure Development ETF (PAVE) are built around companies that manufacture, install, and maintain this physical infrastructure. Their thesis isn't dependent on who sells the power, but on the undeniable need to modernize and expand the delivery system.

Trump's comments also pointed to pressure on PJM Interconnection, the largest grid operator in the U.S., to run emergency auctions for long-term power contracts. This suggests an administration focused on cost control for the broader grid, not on stifling AI growth. The need for action is underscored by stark data: a report from July 2025 noted that capacity prices in PJM for the 2026–2027 period had skyrocketed to $329.17 per megawatt-day, up from just $28.92 two years earlier. That's a signal of serious supply stress.

So, for the investor, the story is evolving. It's no longer just a simple tale of rising demand. It's becoming a story about a massive capital spending cycle. Utilities ETFs might still have a role to play, but not necessarily as sellers of more kilowatt-hours. Their role could be as the builders and owners of the essential infrastructure that delivers power—whether that power comes from their own plants or someone else's.

In the high-stakes race to fuel artificial intelligence, the biggest opportunity may have just shifted. It might not be in who sells the electricity, but in who builds the road it travels on.