California’s utility regulator has mandated procurement of 1 gigawatt of clean firm power by 2026, pulling Cape Station capacity toward both utility and tech buyers

Decision Focus

Google signed a 396-megawatt offtake agreement with Fervo Energy’s Cape Station facility in southwest Utah, with Fervo targeting delivery of 100 megawatts to the grid by end of 2026. One year earlier, Meta backed XGS Energy’s 150-megawatt project in New Mexico. Both deals reflect a hyperscaler push toward clean firm power capable of supporting 24/7 carbon-free energy commitments that solar and wind cannot satisfy without substantial storage. The operational signal underneath both announcements is this: enhanced geothermal — the dominant commercial approach — is a water-intensive technology being deployed in states already stressed by multi-decade drought. Water availability is emerging as a second-order constraint on geothermal scalability that most offtake risk models have not yet priced.

90-Second Brief

Today, google and Meta have committed to a combined 546 megawatts of geothermal capacity across Utah and New Mexico, the two largest hyperscaler geothermal commitments on record. Enhanced geothermal works by fracturing hot dry rock and circulating injected water to generate steam, a method that permanently loses a meaningful share of that water underground. California’s utility regulator has mandated procurement of 1 gigawatt of clean firm power by 2026, pulling Cape Station capacity toward both utility and tech buyers. The intersection of data center cooling demand and geothermal generation demand creates a compounding water exposure that arid Western states are not currently positioned to absorb at gigawatt scale.

What Is Really Happening?

Enhanced geothermal borrows its core technique from oil and gas: hydraulic fracturing creates permeability in hot dry rock, water is injected to capture heat, and that water returns to the surface as steam to spin a turbine. This method expands viable geothermal geography dramatically by eliminating the need for naturally occurring underground reservoirs — but it introduces a structural water loss problem. At Fervo’s first commercial project in Nevada, operators reported losing approximately 30 percent of injected water permanently underground.

The alternative, closed-loop geothermal, addresses water loss directly. XGS Energy’s system circulates the same water continuously through a sealed underground pipe, transferring heat without exposing it to rock fractures — requiring near-zero water addition once operating. Stanford University’s geothermal program has noted, however, that closed-loop systems generally produce less power per well with shorter well lifespans, placing a scaling ceiling on output per unit of capital that enhanced geothermal does not face.

The practical result is a technology split: the approach that scales to data center power requirements consumes significant water; the approach that conserves water cannot yet demonstrate it will scale to those requirements at competitive cost. That gap is the central unresolved constraint in the current geothermal buildout.

Why It Matters for Global Heads of Data Center Energy

A 24/7 CFE commitment backed by a geothermal PPA carries an implicit assumption: the plant will operate at contracted output continuously. If water access becomes constrained — through drought, competing agricultural or municipal allocation, or state water rights adjudication — the plant faces an availability exposure, not merely a price exposure. Unlike basis risk on a wind PPA, water-driven curtailment undermines the underlying carbon accounting, not just the economics.

Researchers at Utah FORGE calculated that supplying 2 gigawatts of capacity entirely from enhanced geothermal could consume 42 million gallons of water per day — roughly half the daily water demand of Albuquerque. Portfolio-level deployment across the Mountain West would require water rights acquisition at a scale that has no established market mechanism, regulatory precedent, or permitting pathway in several key jurisdictions.

There is also a compounding demand problem. Data centers in arid climates already draw scrutiny for cooling water consumption. A geothermal-powered campus drawing from the same watershed as its energy source adds two separate water draws to a single stressed system. Community and regulatory opposition has already constrained data center siting across parts of the Southwest; layering geothermal water intensity onto that dynamic sharpens the exposure before the first turbine spins.

Forward View

California’s clean firm power mandate requires 1 gigawatt of procurement by 2026, and one California utility has already committed to 320 megawatts from Cape Station. That obligation is pulling available early-stage geothermal capacity toward regulated utility contracts in the same window that hyperscalers are competing for direct offtake. If multiple demand pools — hyperscalers, California utilities, and other Western load centers — converge on a thin development pipeline, PPA pricing and queue position will tighten faster than new projects can reach commercial operation.

Enterprise buyers beyond the established hyperscalers are beginning to enter the market. Oracle has reportedly proposed developing up to 2 gigawatts of new generation in New Mexico, with geothermal among the options under consideration. The water demand implied by that scale, if realized through enhanced geothermal, would stress New Mexico’s available supply in ways not currently reflected in state permitting timelines or community consultation processes.

Utah FORGE is actively testing methods to reduce water loss at its research site adjacent to Cape Station, including well spacing optimization and revised injection-recovery protocols. Those results will determine whether enhanced geothermal can close the gap between its current water intensity and what regional hydrology can sustain at the scale data center operators require.

What Is Still Uncertain

Fervo projects that Cape Station’s water loss rate will be substantially lower than the 30 percent recorded at its Nevada site — possibly as low as 5 percent under its revised design — but those figures are projected, not yet measured at operational scale in Utah. Subsurface geology remains incompletely characterized until production data is available. XGS Energy’s closed-loop system has not demonstrated commercial-scale output; the Meta-backed 150-megawatt New Mexico project has no confirmed site location and no published construction timeline. The legal basis for water rights acquisition at large-scale enhanced geothermal projects varies by state and is unsettled in several Mountain West jurisdictions. Transmission access and drilling cost variability add further investment risk that data center demand has eased but not resolved.

One Question for Your Team

Does your geothermal PPA due diligence framework include a water rights audit — covering the developer’s current legal access to water, the conditions under which that access could be curtailed, and whether your data center and its energy source draw from the same stressed watershed?

Sources

  • Upr — Utah could see a geothermal energy boom — but it’ll take water to make it happen (Link)