Hyperscalers might regret embracing natural gas if new forecast proves correct

The rapid expansion of AI capabilities is driving an unprecedented demand for computational power, leading hyperscalers to invest heavily in data center infrastructure. Historically, these tech giants have championed renewable energy sources like wind and solar. However, the sheer scale of energy required for advanced AI models has pushed companies like Amazon, Google, Meta, and Microsoft to pivot towards natural gas, a fossil fuel, as a more readily available and seemingly stable power source for their new gigawatt-scale data centers. This strategic shift represents a significant capital expenditure in physical assets and a deeper entanglement with the volatile energy markets, a domain where these tech companies have limited prior experience.

Noreva

Noreva, an energy research firm, has issued a stark warning that challenges the prevailing complacency regarding natural gas prices. According to Peter Gardett, CEO of Noreva, the market has been "lulled into a sense that gas prices can’t go up," but simple arithmetic suggests a much tighter gas market is on the horizon. The firm's forecast indicates that natural gas prices could soar above $10 per million BTUs in certain U.S. hubs, a dramatic increase from current prices ranging between $2 and $4.50. This projection is based on a confluence of factors: the immense demand pull from hyperscalers, a slowdown in the rate of new supply additions, and the increasing connection of the domestic U.S. gas market to the more expensive global liquefied natural gas (LNG) market through rising exports. This shift means that domestic prices will become more susceptible to international market dynamics, which are often higher and more volatile.

Hyperion

Meta's Hyperion data center in Louisiana exemplifies the scale of these new investments, with the company planning a massive 7.5-gigawatt natural gas power plant to support it. This move, mirrored by Microsoft and Google's gigawatt-scale plants in Texas and Amazon's 7.6-gigawatt facility also in Texas, underscores the industry's reliance on natural gas. These locations, particularly West Texas, were initially attractive due to cheap natural gas, often a byproduct of oil extraction with limited local pipeline infrastructure. However, the article notes that new pipelines are now connecting these regions to broader national and international markets, eroding the localized discount. The potential tripling of natural gas prices would significantly impact the operational costs of these "bring your own power" data centers, where fuel can account for half of electricity generation costs. This could lead to higher "token costs" for AI services or force hyperscalers back onto the grid, potentially driving up general electricity prices for consumers.

Henry Hub

The Henry Hub in Louisiana, a key delivery point for natural gas futures contracts, currently prices gas at just under $3 per million BTUs. While futures contracts generally reflect a stable outlook for the foreseeable future, Noreva's analysis suggests this stability is deceptive. The firm argues that years of relatively flat demand and steady supply additions have created a false sense of security. However, the new dynamic of massive AI demand combined with increasing LNG exports is fundamentally altering the supply-demand balance. The growing connection between the domestic and global gas markets means that even modest price swings near hyperscalers' data centers could be magnified elsewhere, leading to significant regional price differentials. These differentials are precisely what Noreva expects will drive prices in some areas above $10 per million BTUs for extended periods, creating substantial financial headwinds for hyperscalers and potentially fueling consumer backlash over rising utility bills.