By Teresa Rivas
Artificial intelligence is power hungry, and companies that help feed it will be the winners of the new AI cycle.
AI's transformative potential means that even strategists who worry about a bubble believe it won't burst any time soon. But there are concerns different stocks will start to dominate.
Utilities are a good example -- the normally staid, defensive sector soared in 2025, as investors believe that it will benefit from AI's insatiable demand for electricity. However, power generation capabilities aren't infinite, and as AI becomes more widespread, there is a greater chance it will outstrip the electricity available.
That lack of juice could jam up the market this year, warns Jordi Visser, head of AI Macro Nexus Research for 22V Research in his latest commentary this weekend.
"AI's binding constraint has moved from GPUs to electric power, gas turbines, transformers, and grid capacity," he says. "We are at the point where all these bottlenecks are going to impact investors."
The physical infrastructure of the power grid, from transformers to power plants, along with the labor that maintains and runs them, can't keep up with demand from AI. That lack of power is likely to be what delays data center growth, and in turn, is a risk to hyperscaler stocks.
In other words, AI will continue to be adopted by more people and companies, but that won't necessarily benefit the hyperscalers -- they will find themselves increasingly hamstrung by electricity constraints.
As Visser notes, the fact that Nvidia is making more efficient chips exacerbates the issue because they gobble up even more power, and will prolong the bottleneck -- perhaps through the next decade.
"The grid can't give everybody what [they] need as quickly as they want, " Visser says.
He doesn't doubt that data center plays such as GE Vernova are going to make money. But he does question whether, given these constraints, they will be able to surprise on the upside -- and beating market expectations is vital for companies sporting very high multiples.
Therefore, he sees more potential benefit for the companies that can help alleviate this problem, from big energy companies such as Exxon Mobil and Chevron, to natural gas companies and commodity plays.
"Commodities, particularly copper, silver, rare earths, and energy are price-inelastic inputs to AI infrastructure," Visser notes, while countries -- concerned with national security and competition -- will pay almost any price to utilize them.
Likewise the companies that need them aren't constrained by price. That leads him to believe that "energy and basic materials will be the two best performing sectors this year."
Ultimately, it boils down to the fact that having super efficient and powerful chips means nothing if you don't have the electricity to use them.
AI growth pitted against power availability is a "perfect storm" for 2026, as Visser puts it. Investors will have to learn how to ride the wave.
Write to Teresa Rivas at teresa.rivas@barrons.com
This content was created by Barron's, which is operated by Dow Jones & Co. Barron's is published independently from Dow Jones Newswires and The Wall Street Journal.
(END) Dow Jones Newswires
January 12, 2026 15:22 ET (20:22 GMT)
Copyright (c) 2026 Dow Jones & Company, Inc.
