Hyperscalers might regret embracing natural gas if new forecast proves correct
Natural gas prices could triple in some parts of the U.S., which could saddle hyperscalers with massive bills to power their AI data centers.
The hyperscalers' pivot to natural gas was always a hedge against the intermittency of renewables. The logic was simple: when the wind stops and the sun sets, gas turbines keep the AI training clusters humming. But a new forecast suggests that hedge may carry a price tag the industry didn't fully price in.
According to the report, natural gas prices could triple in certain U.S. regions. For companies like Amazon, Google, Microsoft, and Meta, which have signed long-term power purchase agreements tied to gas, that is not a line-item nuisance. It is a structural shift in their operating costs, potentially adding billions to the annual energy bills that underpin their AI ambitions.
The irony is that these same companies have made aggressive public commitments to carbon neutrality. Natural gas was sold internally and externally as a bridge fuel, a pragmatic step toward a cleaner grid. But a tripling in price does not just strain balance sheets; it undermines the narrative that gas is the cheap, reliable partner to renewables. When the bridge toll spikes, the destination starts to look further away.
What makes this forecast particularly pointed is the timing. AI infrastructure buildouts are not flexible. Once a data center is designed around gas turbines, retrofitting for other power sources is a multi-year, capital-intensive project. The hyperscalers are not just buying energy; they are locking in a fuel dependency that could outlast the current price cycle.
There is also a competitive dimension. Not all hyperscalers are equally exposed. Those with more diversified energy portfolios, including nuclear or long-term renewable contracts, may weather a gas spike better than peers who went all-in on the bridge. The market will notice the difference in future earnings calls.
For now, the forecast is just a forecast. But the underlying signal is clear: the era of cheap, abundant natural gas as a silent partner to AI growth may be ending. The hyperscalers made a calculated bet, and the odds may have just shifted against them.