A significant increase in power consumption by artificial intelligence (AI) data centres could dramatically increase the demand for natural gas in the latter half of this decade, according to a new report by Tudor Pickering Holt & Co.
The investment banking firm predicts that up to 8.5 billion cubic feet per day (bcfd) of natural gas could be additionally required to meet this burgeoning demand.
As technology companies continue to expand their use of power-intensive generative AI, concerns are mounting over the capability of the US electrical grid to keep pace. The report highlighted a scenario where US power and technology companies have been compelled to either forge direct agreements with energy producers or to develop their own power supply systems, in an effort to secure sufficient energy resources.
The increasing overall demand for power has significantly contributed to a national backlog of energy projects, which has risen to 2,600 gigawatts of requests in 2023 from 2,000 gigawatts in 2022, as per data from Lawrence Berkeley National Laboratory.
This trend presents a potential boon for natural gas producers and pipeline operators. Companies like Kinder Morgan, Williams, and Energy Transfer, along with gas producers such as EQT and Chesapeake Energy, are well-positioned to capitalize on the growing demand for natural gas.
The report further forecasted that natural gas prices, which reached a three-and-a-half-year low of USD 1.61 per million British thermal units (mmbtu) in February due to mild winter weather, are expected to average around USD 4 per mmbtu in the second half of the decade. This anticipated stabilisation in prices comes after a period of low rates and reduced production, potentially shifting the economic landscape for producers and suppliers.
With AI data center power demand currently at 11 gigawatts, projections suggest an increase to 42 gigawatts by 2030 under the base case scenario.
