While Nvidia and Palantir grabbed the spotlight in the early days of the AI revolution, a more fundamental story is now taking shape: powering the infrastructure behind it. As data centers proliferate globally, electricity is quietly replacing computing power as the hidden fuel determining how far the AI race can go. Asset managers including BlackRock have made it clear that companies supplying electricity to data centers may be the biggest beneficiaries in the AI supply chain.
According to Deloitte, electricity consumption by U.S. AI data centers is expected to surge 30-fold between 2024 and 2035. The International Energy Agency projects that global data center power demand will double to roughly 945 terawatt-hours by 2030. Facing such a staggering gap, the question is which energy sources will lead the charge.
Natural gas plants can be deployed quickly and offer cost-effective, reliable baseload power, making them the go-to solution for immediate capacity needs.
National Fuel Gas Co. (NFG) operates across the Appalachian Basin, covering natural gas extraction, transportation, and sales. The company is actively positioning itself as a key partner for AI data center power demand, with a pipeline expansion project set to deliver gas directly to a data center campus near Pittsburgh.
Archrock Inc. (AROC) provides natural gas compression services, a critical step in moving gas from wells to processing facilities. Its CEO recently noted that AI data center power demand, along with LNG exports, is driving sustained growth, and the company is expanding capacity through capital investment to meet customer needs.
With net-zero commitments from tech giants like Google and Microsoft, nuclear power has emerged as a strategic priority for the decade ahead.
Constellation Energy (CEG) is restarting a reactor at Pennsylvania’s Three Mile Island nuclear plant specifically to supply power to a Microsoft data center nearby—a move that signals the strong return of nuclear energy. Constellation is already the largest nuclear power producer in the U.S., generating more electricity from nuclear fission than all other U.S. utilities combined. Analysts see further upside, with a one-year price target more than 25% above current levels.
GE Vernova (GEV) , the power arm of the former General Electric, spans nuclear reactors, gas turbines, grid optimization, and more. In 2025, total orders jumped to $59.3 billion, pushing its backlog to a staggering $150 billion—roughly four years of work. The company is a clear indirect beneficiary of the strain AI data centers are placing on power grids.
NextEra Energy Inc. (NEE) derives more than half of its generation capacity from renewables, while also holding natural gas and nuclear assets. The company plans to add 30 gigawatts of capacity by 2035 and is partnering with Google to develop data center campuses, using its diverse energy mix to meet AI power demands flexibly.
Powell Industries Inc. (POWL) manufactures electrical equipment for high-value projects in data center end markets. The company carries a $1.4 billion backlog and maintains a strong balance sheet with virtually no debt.
Array Technologies Inc. (ARRY) produces solar tracking systems that help photovoltaic arrays maximize electricity generation. Revenue grew 40% in 2025, though shares have pulled back on a softer outlook, leaving them trading at roughly 10 times forward earnings—a notable discount to peers.
Energy Vault Holdings Inc. (NRGV) focuses on utility-scale energy storage, using sodium-ion batteries and software to manage the highly volatile power loads created by AI training and inference, addressing the intermittency of renewable energy.
The AI energy landscape offers a tiered set of opportunities:
In the AI era, electricity is no longer just infrastructure—it is fast becoming a core variable shaping the industry’s competitive landscape.
