Fervo Locks In 396-MW Google Deal as Hyperscalers Race for Firm Geothermal Power
The largest-ever geothermal PPA signals that US data center operators are willing to bet on unproven supply at scale to secure round-the-clock clean power.
Fervo Energy signed a 396-megawatt power purchase agreement with Google on Tuesday (2026-09-01), the largest geothermal PPA ever struck, with Google describing the output as a "foundational building block" for a potential data center in southwest Utah.8,7
The deal's scale matters less as a geothermal milestone than as a statement about what hyperscalers can no longer find easily: dispatchable, around-the-clock clean electricity. Wind and solar have flooded many US grids, but they cannot guarantee power when a data centre demands it at 2 a.m. on a still night. Geothermal can.8,7
Fervo's commercial model rests on adapting horizontal drilling and fracking techniques — skills the US shale industry developed over two decades — to heat-exchange loops drilled into hot rock. The company plans to deploy what it calls standardised 50-MW GeoBlocks that can be aggregated into larger GeoClusters, a modular approach designed to make permitting and construction repeatable rather than bespoke. William Blair analysts flagged, however, that repeatability remains the critical unproven variable: Fervo needs more project milestones before investors can confidently model a replication curve.8,6
Context helps here. The US currently has roughly 4 gigawatts of installed geothermal capacity, and Fervo alone has 3.65 GW in construction, ready-to-build, or advanced development stages — a potential near-doubling of the national fleet from a single company. Less than 1% of global electricity came from geothermal as of May 2026, according to The Economist. Princeton University researchers estimate that technical advances could, by 2050, allow geothermal to produce nearly triple the current output of America's nuclear fleet, which supplies around 20% of national electricity.2,1
Capital is beginning to move toward that thesis. The International Energy Agency forecasts cumulative global geothermal investment could reach $1 trillion by 2035, against the $1 billion to $2 billion invested annually in 2024. That is a dramatic implied acceleration, and most of it is contingent on costs falling as the drilling learning curve steepens.1
The demand pull is not subtle. US data centre power consumption is expected to rise nearly 360% between now and 2030 to reach 110 GW, according to industry forecasts, as AI workloads scale. That trajectory is forcing large technology companies to secure long-term baseload contracts years before projects come online. Google has contracted more than 22 GW of clean energy since 2010 across a range of instruments, including hydropower, solar, demand response, and now enhanced geothermal.4,3
Fervo is not alone in attracting attention. Another next-generation geothermal developer, whose name The Economist reported in May 2026, was valued at roughly $1.4 billion and expected to begin producing electricity under a 500-MW deal with Shell's power division and a Californian utility. That project's first phase was targeted to come online in 2027. Two developers cited by OilPrice.com in July 2026 reported drilling updates suggesting their costs were beginning to undercut nuclear power on a levelised basis, though those comparisons carry significant methodology caveats.1,5
The Trump administration has been pushing geothermal alongside other domestic energy additions under executive orders targeting grid reliability, and the Utah desert setting of Fervo's project sits within what geologists consider one of the more accessible heat resources in the continental US. Iceland and Kenya hold more concentrated geothermal hotspots globally, but the US combination of open land and proven fracking expertise gives domestic developers a structural advantage in deploying enhanced geothermal at scale.6
Google's broader procurement strategy illustrates the arms race underway. Beyond the Fervo deal, the company reached 1 GW of demand-response capacity under long-term contracts with utilities including the Tennessee Valley Authority, DTE Energy, and Minnesota Power, allowing it to curtail or shift compute workloads during grid stress periods. It also signed a hydropower framework with Brookfield covering up to 3 GW nationally. The Utah geothermal PPA fits a pattern of locking in every available form of firm or flexible clean power, not just the cheapest.3
Whether Fervo can deliver 396 MW on schedule and at the contracted price will be answered incrementally as its GeoBlock deployments progress. Drilling costs are the central variable: if the company can replicate the per-well economics of its existing Utah operations across new sites, the model works. If it cannot, the deal joins a longer list of clean power agreements that looked compelling at signing but stalled on execution. The next concrete signal will be Fervo's drilling performance data from its initial GeoCluster phases — that is where the repeatability case either holds or begins to crack.8,6,2