Hyperscalers might regret embracing natural gas if new forecast proves correct

Leading technology companies such as Amazon, Google, Meta, and Microsoft are increasingly relying on natural gas to fuel their AI data centers, despite potential risks. A research report from Noreva, an energy research firm, warns that natural gas prices in the U.S. could surge by up to three times within years due to a confluence of declining supply growth, rising exports of liquefied natural gas, and hyperscaler demand.

CEO Peter Gardett cautioned that hyperscalers may not be prepared for such price shocks. He noted that companies like Meta, Microsoft, and Google have already built large-scale natural gas power plants in Texas and Louisiana to meet their needs. Amazon is also planning a similar plant in Texas. While prices for natural gas currently hover around $2 to $4.50 per million BTUs, Noreva predicts they could rise above $10 in certain regions.

This increase would significantly impact the cost of running these AI data centers, potentially leading to higher token prices or encouraging hyperscalers to connect to the grid. The stability of natural gas prices is largely due to a combination of stable demand and new supply additions. However, Gardett believes that energy companies will struggle to add new supply at the same rate as before.

The recent connection of the domestic gas market to global gas markets and the surge in AI demand are key factors driving up prices. West Texas, where many of these companies have located their plants, has traditionally produced large amounts of natural gas as a byproduct of oil extraction. However, new pipelines are now connecting this region to national and international markets, leading to increased demand and potential price differentials.

This could result in extended periods of elevated natural gas prices in some regions, potentially impacting hyperscalers' operations and driving up their energy costs.