USGS estimates 135 GW of enhanced geothermal potential in the Great Basin1; OilPrice.com’s April 25 analysis puts total US capacity at up to 150 GW1 (unverified by independent third parties as of this writing). Current US conventional geothermal stands at roughly 2.7 GW across 99 plants1, concentrated in California and Nevada. AI data center procurement teams competing over a thin inventory of nuclear restart PPAs and SMR contracts now have a third firm-clean option to price against either.
How the Capacity Estimate Changes Procurement Math
The 150 GW figure requires sourcing scrutiny up front1. OilPrice.com’s April 25 piece attributes 135 GW to USGS estimates for the Great Basin specifically; the broader 150 GW total-US figure1 comes from unspecified “other predictions” without traceable attribution. Use the 135 GW number for anything load-bearing1.
That said, 135 GW dwarfs what’s deployed1. The US runs about 2.7 GW of conventional geothermal today, all of it hydrothermal reservoirs at commercially accessible locations. Enhanced Geothermal Systems (EGS) work differently: drill into dry hot rock, fracture it hydraulically, inject water, extract the heat. The resource base expands dramatically because it no longer depends on naturally occurring permeable rock or subsurface water. The constraint has always been the drilling technology, which is what makes Cape Station relevant.
The Department of Energy committed $171.5 million2 in February 2026 for geothermal field-scale tests and exploration drilling. That’s a policy signal that the technology gap is considered closeable rather than fundamental.
For hyperscalers that have spent two years locking up nuclear restart PPAs, EGS offers something nuclear can’t easily replicate: greenfield development without an NRC operating license. The NRC licensing track runs on multiyear timescales and can be derailed by policy changes, public opposition, or grid interconnection disputes that have nothing to do with the plant itself. EGS routes around that constraint, trading it for a different risk register that runs through state agencies and the Bureau of Land Management.
Cape Station: The EGS Proof Point
Fervo Energy’s Cape Station is planned at 500 MW total in Beaver County, Utah1. The first 100 MW phase is under development and expected to come online later in 20261; that phase alone would make it the world’s largest EGS facility. The remaining 400 MW follow in subsequent phases1.
Fervo’s technique borrows from oil and gas: horizontal drilling and multi-zone hydraulic fracturing applied to geothermal rock rather than hydrocarbon-bearing formations. The technology transfer is deliberate. The Western US has an established workforce trained in those techniques and an existing equipment supply chain. Fervo has leased nearly 600,000 acres across the Western US1, with stated potential to develop over 42 GW from that land position1.
On April 17, 2026, Fervo filed a registration statement for an IPO3, planning to list Class A shares on Nasdaq under ticker FRVO. The timing is pointed: pairing Cape Station’s proof point with a public capital raise asks markets to price Fervo’s 42 GW land position against the risk that EGS doesn’t perform at utility scale. If the offering prices cleanly, it validates the thesis in a way DOE grant announcements can’t. Capital markets have money at risk; DOE program offices don’t.
Geothermal vs. Nuclear and SMRs for Data Centers
For a procurement team evaluating firm clean baseload today, the three options carry distinct risk profiles:
| Option | Licensing path | Scale potential | Key risk |
|---|---|---|---|
| Nuclear restart | NRC license required | Site-limited, GW-scale where viable | Regulatory timeline, overrun history |
| SMRs | NRC design + site license | No commercial US precedent | First-of-kind construction uncertainty |
| EGS | State/BLM permits, no NRC | USGS estimate (Western US)1 | Seismicity, water use, subsurface lease |
The NRC comparison is the clearest argument for EGS. Combined construction and operating licenses for new nuclear units have historically overrun their initial schedules by years; the restart track is faster but depends on a finite inventory of plants in recoverable condition. SMRs have no commercial operating precedent in the US and face the same NRC path as conventional nuclear plus first-of-kind construction risk on top. A procurement team trying to sign a deal with a 2027-2028 delivery target can have a rational conversation about BLM permitting timelines in a way it cannot about NRC review timelines for novel reactor designs.
EGS doesn’t escape licensing. BLM geothermal leases on federal land require environmental review under NEPA, and state-level permitting for water use and induced seismicity is real. But the failure modes are different from the NRC path, and the oil-and-gas permitting precedent library is extensive.
Dispatchability is the other axis. Wind and solar continue to fall in cost, but both are intermittent. Firm clean power commands a premium precisely because it doesn’t require paired storage. Geothermal runs at high capacity factors year-round; the load profile for a hyperscale data center (continuous, non-seasonal, indifferent to whether it’s July or January) matches the generation profile well. That match is worth paying for if the alternative is buying dispatchable firm capacity elsewhere on the grid at peak prices.
The New Binding Constraint: Subsurface Leases
A 135 GW Great Basin resource estimate1 shifts where the competitive bottleneck in data center site selection actually sits. Current site-selection calculus centers on transmission: interconnection queue position, available capacity, proximity to fiber and customers. Northern Virginia’s data center concentration reflects interconnection availability as much as anything else. The geothermal resource is not there.
EGS potential is concentrated in Utah, Nevada, Idaho, and adjacent Western states where the geothermal gradient is high enough to make drilling economics work. A campus sited to take direct geothermal supply is in Beaver County, Utah. That’s not inherently prohibitive. Hyperscale buildouts are already moving into the Mountain West for power cost reasons, which means the competitive constraint shifts from queue position in PJM to subsurface lease availability in BLM geothermal priority zones.
Fervo’s 600,000-acre Western US land position1 reads differently when framed as a site selection moat rather than a balance-sheet line item. If Cape Station validates EGS at scale, that land position becomes a constraint on how fast anyone else can build. BLM geothermal lease availability in high-gradient corridors is finite; the federal leasing process runs on its own calendar. Developers who wait for the Cape Station proof point before acquiring acreage will find the obvious positions already taken.
One clarification on the transmission argument: EGS plants still require grid interconnection. Geothermal doesn’t bypass interconnection queues the way a behind-the-meter arrangement does. The structural advantage is different: a large data center collocated with a geothermal plant can negotiate a single large interconnection point rather than competing for incremental capacity in a congested regional queue. Several hyperscale campuses have structured nuclear co-location deals on similar logic.
Risks: Seismicity, Water Use, and Permitting
EGS carries two risks that appear in every serious engineering review and have terminated demonstration projects before Cape Station reached commercial scale.
Induced seismicity is the first. The hydraulic fracturing that opens dry hot rock also produces microseismic events. Most fall below felt threshold, but projects in Basel and Pohang were suspended after earthquakes in the Mw 3.0+ range triggered regulatory action and public opposition. Fervo’s earlier Nevada pilots produced a manageable seismicity profile by published accounts, and Cape Station’s Beaver County geology differs from those European demonstration sites. The 100 MW first phase1 will generate the dataset that either confirms the favorable profile or complicates it at commercial scale.
Water use is the second. EGS injects water to extract heat; closed-loop designs recycle most of it, but losses require makeup water. The Great Basin is not water-rich. State water rights in Nevada, Utah, and Idaho are already contested among agricultural, industrial, and municipal users. A geothermal project requiring significant water allocation enters that dispute as a new party, which is a permitting vector opponents will use. Whether this is a project-stopper or a project-complicating factor depends on local hydrology and state water law. There is no uniform answer across the Western US.
The BLM permitting path adds NEPA environmental review for federal land projects, measured in months to a few years rather than a decade, but not automatic. The cumulative permitting picture is lower regulatory risk than nuclear, but different enough that nuclear procurement teams shouldn’t assume their existing risk frameworks translate directly.
What Cape Station tests over the next 12 to 18 months is whether EGS can simultaneously clear the technology bar and the permitting bar at commercial scale. The USGS resource estimate establishes the resource case. The DOE’s $171.5 million commitment2 establishes the policy backing. The Fervo IPO filing3 establishes a financing model. The first 100 MW in Beaver County1 is the test of whether any of that arithmetic is real.
Frequently Asked Questions
What capacity factor can a procurement team bank on from an EGS plant versus nuclear or solar?
Conventional geothermal delivers 90%+ annual capacity factors—comparable to nuclear baseload and roughly triple utility-scale solar (~25%) or offshore wind (~45%). EGS hasn’t accumulated enough run-time at commercial scale to confirm that figure, but the thermodynamics of subsurface heat exchange don’t change with the drilling method; the risk is thermal drawdown over decades, not intermittency.
Does an EGS reservoir lose output over time the way a gas field depletes?
Yes. Injected water gradually cools the fractured rock over a 20-50 year horizon, reducing heat extraction unless the operator drills laterally into fresh zones or hydraulically fractures new intervals. That decline curve is a PPA structuring problem unique to geothermal—nuclear fuel contracts and solar irradiance don’t degrade on the same timeline, so buyers accustomed to those frameworks need to model thermal depletion explicitly.
Could EGS serve data centers in the Eastern US, or is this strictly a Western resource?
Technically EGS works anywhere if you drill deep enough, but the geothermal gradient east of the Rockies is roughly half the Great Basin’s, pushing target depths to 5-8 km versus 2-4 km in Utah and Nevada. That roughly doubles per-well drilling cost, which is why the USGS 135 GW estimate concentrates the economic resource in Western states. Eastern data center clusters in Northern Virginia would need to procure EGS power through long-distance transmission or pair it with other firm sources.
How long does it take to acquire a new BLM geothermal lease if a developer doesn’t already hold acreage?
BLM geothermal leasing is nomination-driven, not calendar-driven: parcels reach a competitive sale only after industry nominations clear NEPA review, and recent Nevada lease sales have run roughly annually depending on BLM staffing and environmental workload. Fervo’s early 600,000-acre position across the Western US is difficult to replicate quickly because the nomination-to-award cycle can stretch 18-36 months for parcels in high-gradient corridors that haven’t been previously leased.