Hyperscalers might regret embracing natural gas if new forecast proves correct
Major technology companies are increasingly planning to use natural gas to power the energy-intensive data centres supporting artificial intelligence, but a new forecast warns this could expose them to steep fuel-price rises. The report matters because higher gas costs could make privately powered data centres far more expensive to operate, potentially increasing AI service prices or placing additional pressure on electricity grids.
Energy research firm Noreva forecasts that gas prices at some US hubs could exceed $10 per million BTUs, compared with roughly $2–$4.50 today and just under $3 at Henry Hub. Meta, Microsoft, Google and Amazon have recently announced gas-powered projects ranging from gigawatt-scale plants to 7.6 GW, largely in Texas and Louisiana; Noreva says slower supply growth, more costly new wells, LNG exports and AI-driven demand could tighten the market despite currently stable futures prices.
- AI data-centre gas plans may face major price risks.
- Noreva predicts some US gas prices could more than triple.
- Exports and rising demand could tighten supply.
New here? Start with this
Artificial intelligence systems are run in large data centres packed with powerful computer equipment. These sites use vast amounts of electricity, and technology companies are seeking reliable power supplies as demand for AI services grows.
Natural gas is a fuel used to generate electricity, either through the main grid or in power plants built for particular facilities. Meta, Microsoft, Google and Amazon are among the companies considering or developing gas-based power projects, especially in parts of the southern United States where new data centres are being built.
Gas prices can change with drilling levels, demand from households and industry, and exports of liquefied natural gas overseas. If fuel becomes more expensive, the cost of running gas-powered data centres could rise, affecting company spending and adding to wider competition for electricity supplies.
Both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
The forecast highlights a genuine commercial and system-wide risk: large, long-lived gas commitments could leave hyperscalers exposed if supply growth disappoints while LNG exports and new data-centre demand rise together. Higher and more volatile fuel costs would weaken the economics of privately generated power, potentially feeding through to AI prices and competing with other consumers for constrained gas and electricity infrastructure. Supporters of this warning argue that companies with substantial resources should favour efficiency, renewables, storage and cleaner firm-power options rather than lock in another source of fossil-fuel and price risk.
The case against
Natural gas can be a pragmatic response to the immediate need for reliable, dispatchable electricity as data-centre demand grows faster than transmission and clean generation can be built. Companies may be able to manage fuel exposure through long-term contracts, hedging, diversified locations and efficient modern turbines, while gas plants can support grid reliability and complement variable renewable power. A single forecast is necessarily uncertain, and advocates of the projects would argue that delaying capacity on the assumption of sustained high prices could itself impose greater costs through constrained AI services, missed investment and continued reliance on less dependable alternatives.