📰 Key Takeaway

After aggressively snapping up wind and solar projects in recent years, hyperscale cloud providers are now pivoting to bet on natural gas as the power source behind the data centers fueling their AI ambitions. But a new research report suggests this fresh fossil-fuel romance could come back to bite them.

Energy research firm Noreva says that as hyperscalers’ power demand growth slows and clashes with rising LNG export volumes, natural gas prices in parts of the US could triple over the next few years — and hyperscalers may be totally unprepared for the price shock. Noreva CEO Peter Gardett told TechCrunch that everyone in the energy market has been lulled into complacency by the illusion that “gas prices won’t rise,” when some simple arithmetic points to a far tighter gas market than we had just a few years ago.

On the specifics: Meta announced in March it would build a massive 7.5-gigawatt natural gas plant in Louisiana to power its Hyperion data center; days later, Microsoft and Google each announced they’d build their own gigawatt-scale gas plants in Texas; Amazon also plans a 7.6-gigawatt gas plant in Texas. Noreva expects gas prices at some trading hubs to blow past $10 per million BTU, compared to roughly $2-$4.5 today (Henry Hub is currently sitting just under $3). Since fuel costs make up about half the cost of running a large power plant, doubling or tripling gas prices could sharply drive up operating costs for “self-generated power” AI data centers — pushing up token costs, or forcing operators to switch back to the grid and driving up electricity prices there instead.

Gardett also noted that current forward gas futures contracts don’t reflect any major price shift, but he doesn’t think the market has this one right. The reasoning: demand has been flat and new supply has kept offsetting output declines from aging wells, but development costs for new wells are climbing, so supply growth probably can’t keep up its past pace.

💬 JudyAI Lab Take

Hyperscale cloud providers are pivoting hard toward natural gas to meet the power demands of their AI data centers — but a new research report suggests this rush could be blinding them to the risk of a future spike in electricity prices.

Meta, Microsoft, Google, and Amazon have all recently announced plans to build gigawatt-scale natural gas plants in Louisiana and Texas to power their data centers. Noreva CEO Peter Gardett notes that the market widely believes gas prices won’t rise significantly, but as LNG exports grow and the cost of developing new wells climbs, prices at some trading hubs could blow past $10 per million BTU in the next few years — more than triple current levels. Fuel costs account for roughly half of a power plant’s generation costs, meaning operating costs for AI data centers that generate their own power could rise substantially, ultimately showing up in either token pricing or electricity rates. It’s a reminder that AI infrastructure costs aren’t just about chips and power contracts — long-term shifts in energy market supply and demand are just as critical a variable.

Worth keeping an eye on natural gas futures prices and public data center power contracts as a leading indicator for where AI service pricing is headed.

📅 Source Info

🔗 Further Reading