Hyperscalers like Amazon, Google, Meta, and Microsoft, which have been developing wind and solar power for years, are betting that natural gas can power the data centers behind their lofty AI ambitions. But a new research report suggests they may be regretting their newfound affinity for fossil fuels.
Natural gas prices could triple in some parts of the U.S. over the next few years as hyperscaler demand collides with lower supply growth and increased liquefied natural gas exports, according to energy research firm Noreva. Hyperscalers may not be prepared for future price shocks.
“I think everyone in the energy market has settled into the feeling that gas prices are not going to go up,” Noreva CEO Peter Gardet told TechCrunch. “It takes some simple math to get to a gas market that is much tighter than it was just a few years ago.”
Hyperscalers’ big bet
Due to low gasoline prices, hyperscalers are locking in a portion of the market. Meta announced in March that it would build a massive 7.5 gigawatt natural gas power plant in Louisiana to power Hyperion’s data centers. Days later, Microsoft and Google each announced they would build their own gigawatt gas power plants in Texas. And Amazon isn’t left behind, with plans to build a 7.6 gigawatt gas power plant in Texas.
For companies that had traditionally avoided large capital investments, the data center construction boom suddenly forced them to invest heavily in the physical world and move deeper into unfamiliar energy markets.
Gardette said at least one investor he spoke to was “surprised” by how much natural gas price risk hyperscalers were willing to take on. “They’re doing things that aren’t normal for off-takers,” he said.
Noreva expects natural gas prices to soar above $10 per million BTU at certain hubs and delivery points for futures contracts. Currently, prices range from about $2 to $4.50 per million BTUs, with Henry Hub, widely traded in Louisiana, costing just under $3.
Fuel represents about half the cost of electricity for large power plants, so doubling or tripling the price of natural gas could make it even more expensive to operate a “captive-powered” AI data center. That could raise the cost of the token or encourage hyperscalers to connect to the grid, raising power prices.
Natural gas prices appear to be stable in the near term, with futures contracts not expecting major fluctuations. “It’s not an unreasonable bet,” Gardette said. But he’s not convinced they’re right.
Rapid increase in demand
Gardette said natural gas prices have remained stable thanks to years of relatively flat demand and the steady addition of new supply to offset declining production from older wells. He expects energy companies will be able to add more supply, although not as quickly as before. Additionally, new wells are more expensive.
“That alone is not going to change the economic situation here. What is changing the numbers is that we are finally connecting the domestic gas market to the global gas market,” he said. “And the second thing is AI driving demand.”
Hyperscalers are flocking to Texas and Louisiana, drawn by low natural gas prices. Particularly in West Texas, most wells are focused on oil, and the natural gas that comes out of them is a byproduct that doesn’t have much of a market. There weren’t many large pipelines to move natural gas out of the area, so producers sold it at a discount to anyone who could use it. But that is changing.
“They’re finally building some pipelines, many of which are going to export markets,” Gardette said.
As West Texas becomes more connected to domestic and international natural gas markets, demand there will influence prices elsewhere, and vice versa. Even a small price change near a hyperscaler’s large data center can be magnified elsewhere.
“Next to a place where you can get a lot of gas, there’s a place where you don’t have gas, and that makes a huge difference,” Gardette said. This difference can result in prices exceeding $10 per million BTU in some regions over time.
In that scenario, even if hyperscalers are able to withstand higher prices, natural gas consumption could add another dimension to the data center rebound. 80% of consumers are already concerned about the impact data centers will have on their utility bills, primarily electricity-related. That concern could spill over into natural gas prices.
In their rush to power AI data centers, hyperscalers are rapidly entering the fossil fuel world. This is an area in which they have relatively little experience, but it can quickly make a significant impact on your business.
“In the upcoming Alphabet earnings call, we’re going to be talking about the correlation between natural gas prices and Google results, which is strange, but that’s our position,” Gardette said.
If you buy through links in our articles, we may earn a small commission. This does not affect editorial independence.
