Hyperscalers might regret embracing natural gas if new forecast proves correct

Major tech companies are shifting from renewables to natural gas for their AI data centers, investing heavily in gas plants. This pivot opens them to major price swings and new energy market risks. A fresh forecast suggests natural gas prices could triple, endangering the financial stability of their AI projects.

Key takeaways

• Hyperscalers (Amazon, Google, Meta, Microsoft) are investing in gigawatt-scale natural gas plants for AI data centers.
• Energy research forecasts natural gas prices could triple, soaring above $10 per million BTUs in some U.S. regions.
• Higher gas prices could make AI data centers much more expensive to run, potentially driving up AI token costs.
• The move ties hyperscalers’ financial results to volatile fossil fuel markets, a new and risky exposure.
• Consumer anxiety over data centers’ impact on utility bills could extend to natural gas costs.

The Big Bet: Hyperscalers Locking In Natural Gas for AI

Following years of championing wind and solar, Amazon, Google, Meta, and Microsoft now place huge wagers on natural gas for their AI data centers. This represents a clear strategic turn, committing these firms to large capital projects and new energy market territory.

Major Gas Plant Investments

Meta intends to construct a 7.5-gigawatt natural gas power plant in Louisiana. Microsoft and Google each plan gigawatt-scale gas plants in Texas, and Amazon aims for a 7.6-gigawatt plant also in Texas.

Embracing Price Risk

Peter Gardett, CEO of energy research firm Noreva, mentioned that at least one investor expressed surprise at the scale of natural gas price risk these companies accept. He observed, “They’re doing things that are not normal for an off-taker to do.” Historically cautious about big capital spending, the data center surge now pulls them into physical infrastructure and deeper energy dealings.

The Price Shock Forecast: Tripling Costs and Market Tightening

Noreva predicts natural gas prices may triple in some U.S. areas, jumping above $10 per million BTUs from today’s $2 to $4.50 range. CEO Peter Gardett says the market feels “lulled into a sense that gas prices can’t go up.” He expects a supply squeeze as hyperscaler demand meets slower supply growth and rising LNG exports.

Impact on AI Data Center Economics

Fuel makes up roughly half the expense for electricity from a big power plant. So, a doubling or tripling of natural gas prices could drastically raise operating costs for the “bring your own power” AI data centers these companies build. Higher expenses might then increase AI token costs or force hyperscalers onto the grid, which would push electricity prices up for everyone.

Current futures markets show stability and don’t predict big shifts, but Gardett doubts their accuracy. He highlights two critical changes: the link between domestic and global gas markets and the fresh “AI demand pull” from hyperscalers. New pipelines moving gas from areas like West Texas to export markets could create severe local price differences, keeping prices above $10 in some regions for long stretches.

The Demand-Pull Perfect Storm: AI and Global Markets

Energy research firm Noreva identifies two forces reshaping natural gas markets. First is the connection of the U.S. gas market to global trade through LNG exports. Second is the new demand surge from AI data centers.

Hyperscalers originally moved to Texas and Louisiana for cheap natural gas. In West Texas, this gas was often a byproduct of oil drilling with few pipeline options, so producers sold it cheaply. That dynamic is shifting as new pipelines get built, sending much of that gas to export markets.

West Texas integration into wider markets means local data center demand will affect broader prices, and global shifts will impact local costs. This connection can amplify small price movements. Peter Gardett of Noreva says it will create areas with large price gaps, where costs could stay above $10 per million BTUs for long periods, compared to the current $2 to $4.50.

Business and Backlash Risks: From Earnings to Consumer Bills

This shift embeds hyperscalers in the fossil fuel sector, an area where they lack deep experience but which may soon affect their earnings directly. Noreva CEO Peter Gardett forecasts that future Alphabet earnings calls will include analysis linking natural gas prices to Google’s performance. “That is strange, but that’s where we are,” Gardett stated.

Financial Exposure and Consumer Anxiety

Noreva anticipates natural gas prices climbing above $10 per million BTUs in specific hubs, a sharp rise from today’s $2 to $4.50. Because fuel accounts for about half of a big power plant’s electricity cost, a price doubling or tripling could make operating AI data centers far pricier. This might increase token costs or pressure companies to grid-connect, raising electricity prices further.

Beyond pure cost, hyperscalers’ natural gas use could intensify the existing data center backlash. Already, 80% of consumers express concern about data centers affecting their utility bills, mainly regarding electricity. The report indicates this worry may soon include natural gas expenses, too.