The artificial intelligence trade in the US stock market is moving beyond chipmakers. Investors who first chased Nvidia, AMD and other semiconductor names are now looking at companies that can supply the electricity, grid equipment and cooling systems needed to run AI data centres.
AI models need huge computing power, which sits inside data centres, who in turn need electricity around the clock. As a result, power producers, grid contractors and electrical equipment makers have become part of the AI investment story.
US power demand is expected to hit fresh records in 2026 and 2027 as AI use, data centres and electrification increase electricity consumption, according to the US Energy Information Administration. Reuters reported that US power use is projected to rise from a record 4,195 billion kilowatt-hours in 2025 to 4,271 billion kWh in 2026 and 4,397 billion kWh in 2027.
That demand is already creating stress. PJM Interconnection, the largest US grid operator, recently issued warnings as high summer demand pushed up electricity prices and strained transmission lines, especially in northern Virginia, one of the world’s biggest data-centre hubs. Reuters reported that spot power prices in parts of PJM jumped from around $30 to more than $300 per megawatt-hour during peak demand.
Why power is the new AI trade
For Wall Street, the power story has become a second leg of the AI boom. The first leg was chips. The next is the physical infrastructure required to make AI work.
The International Energy Agency estimates that data centres consumed about 415 terawatt-hours of electricity in 2024, or roughly 1.5% of global electricity consumption. It said data-centre power use has grown at 12% a year over the past five years.
Goldman Sachs Research has projected that global data-centre power demand could rise 50% by 2027 and as much as 165% by the end of the decade from 2023 levels. This is why stocks such as Constellation Energy, Vistra, GE Vernova, Eaton, Vertiv, Schneider Electric and Quanta Services are now being tracked alongside AI chip names. Each company gives investors a different way to play the same theme.
Also Read: Abbott Labs shares jump 12% as Q2 sales rise, profit outlook raised
Constellation and Vistra draw nuclear and power demand bets
Constellation Energy and Vistra are among the clearest power-generation plays. Both have gained attention because large technology companies need reliable power for data centres. Nuclear and gas-fired plants are attractive because they can supply power continuously, unlike wind and solar, which depend on weather.
Constellation has been central to the nuclear revival story. The company signed a power supply deal with Microsoft in 2024 to help restart a unit of the Three Mile Island nuclear plant in Pennsylvania. Reuters reported that the deal was tied to Microsoft’s data-centre electricity needs.
The company is now working to restart the plant under the name Crane Clean Energy Center. Reuters reported in June that US regulators granted a waiver linked to the restart plan, with Constellation targeting a restart next year. However, the stocks fell over 30% in the year-to-date period.
Vistra has also become a favourite among investors looking for exposure to rising US power prices and tight electricity supply. The company owns a large power-generation fleet, including gas and nuclear assets. Investors see it as a direct beneficiary if data-centre demand lifts electricity prices and keeps power plants running at higher utilisation. Vistra is down about 12% YTD.
GE Vernova, Eaton and Schneider sell the equipment
GE Vernova, which has gained about 60% this year is more about the equipment side of the trade. AI data centres need turbines, transformers, switchgear, electrical distribution systems, automation and grid-control technology.
The surging demand from AI data centres is worsening shortages of grid equipment such as transformers, pushing up costs and forcing utilities and developers to order years in advance.
GE Vernova benefits from demand for gas turbines, grid equipment and power-generation systems. Eaton and Schneider Electric (rose 86% YTD) benefit from electrification, power management and data-centre electrical systems. These companies are less exposed to daily power prices and more tied to capital spending on infrastructure.
Vertiv and Quanta Services ride the buildout
Vertiv is a direct data-centre infrastructure play. It provides power and cooling systems used inside data centres. As AI servers become more power-intensive, cooling and power management become more critical. The shares have gained nearly 70% this year.
Quanta Services, who shares gained 44% this year, is the grid construction and engineering play. It builds and maintains electric transmission and distribution infrastructure. Reuters reported earlier this year that Quanta forecast 2026 profit above Wall Street estimates, helped by AI-infrastructure demand for its electric segment.
For investors, Quanta is a way to bet on the physical grid work required to connect new data centres and upgrade ageing networks.
Risks are also rising
The trade is powerful, but it is no longer cheap or undiscovered. Many of these stocks have already run hard. Investors are also watching whether regulators push back against rising electricity bills. Reuters reported that the White House plans to bring utilities and data-centre developers together around a pledge aimed at preventing AI power costs from being shifted to consumers.
There are also execution risks. Power projects take years. Grid connections are slow. Transformers and turbines are in short supply. Data-centre projects can be delayed if power is unavailable.
(Disclaimer: Recommendations, suggestions, views and opinions given by the experts are their own. These do not represent the views of Economic Times)
