Fervo’s 396 MW Google PPA turns geothermal into an AI-grade “24/7” contract test—without proving the margin math yet

On Sept. 1, Fervo FRVO announced a 396 MW power purchase agreement (PPA) with GOOG tied to its Cape Station GeoCluster in Utah, describing the offtake as “around-the-clock” / “24/7, carbon-free” electricity and saying the facility is expected to come online in 2028. The agreement also includes an option that can expand Google’s offtake by about 600 MW—taking it to nearly 1 GW—by June 2030.

That set of terms matters because it forces geothermal—historically capacity-limited by reservoir uncertainty and project delivery risk—into the same commercial box as “firm” AI power: a contract that hyperscalers can underwrite for reliability, emissions, and scheduling.

Fervo’s Sept. 1 PPA announcement is a demand-and-capacity proof point for its Cape Station build. However, the investment case for an “AI-grade multiple” hinges on repeatability: delivering firm contracted power at a cost curve that lets the project-specific PPA premium survive drilling variability, equipment lead times, and grid-connection constraints.

That’s why investors often react before the accounting does—because market multiples can move on credibility (a signed PPA with a top-tier buyer), while financials lag until commissioning converts contract megawatts into revenue.

The AI-power competition is not just “clean vs. clean.” It’s “firm vs. firm” under the same operational expectations: dispatchability, contractability, and scheduling certainty.

That puts Fervo FRVO in a multi-constraint contest against:

• Dispatchable gas PPAs (price volatility risk, but fast time-to-commission)
• Nuclear (very firm, but long build timelines and different risk allocation)
• Contracted renewables paired with firming or grid access (where grid congestion can become the binding constraint rather than generation).

From a fundamentals perspective, Fervo’s published financials show the company is still early in the value-creation curve: FY2025 revenue is reported at $138M, operating income remains deeply negative, and operating cash flow is also negative.

That means the valuation “translation” investors are making—turning a signed 396 MW PPA into an AI-grade multiple—depends on a belief that future commissioning will convert today’s capital intensity into a higher-margin operating profile, not just larger project headcounts.

Fervo’s latest delivery update (Q2 FY2026 results coverage) says mechanical completion was achieved on GeoBlocks 1 and 2, while mechanical completion for GeoBlock 3 is expected “over the coming months,” along with targeted commissioning/power timelines for first power and initial power windows.

In practice, the next catalyst stack is likely to be: (1) mechanical completion confirmations, (2) first power/initial power, (3) steady-state performance and commercial operations, and (4) any follow-on PPAs that validate that demand isn’t one-off.

If the hyperscaler “24/7” thesis sticks, it doesn’t just reward geothermal developers. It changes contracting dynamics across the supply chain—especially equipment manufacturing, EPC/O&M execution, and grid deliverability.

In public markets, that often shows up as a rotation between:

• Geothermal specialists with operating fleets (more direct leverage to “proven uptime”)
• Grid and regulated utilities (who can either enable firm delivery via upgrades or face cost-recovery friction)
• Dispatchable power incumbents (gas and nuclear compete on firm reliability and price stability).

The market’s early reaction to Fervo FRVO therefore acts like an options market on firm-clean reliability—not unlike how nuclear/gas contract cycles can re-rate winners before earnings.