Fervo's Geothermal Order Book Hits $11.9 Billion on AI Power Demand

Fervo Energy reported $11.9 billion in contracted revenue in its third quarter earnings released Monday, up from $7.2 billion at the end of the previous quarter. The 65% jump in one quarter marks the clearest signal yet that hyperscalers, the largest operators of AI data centers, are willing to sign multi-year power contracts with a geothermal startup that only connected its first commercial-scale plant to the grid earlier this year.

The company is still losing money. Fervo posted an operating loss of $43.7 million for the quarter. It holds $1.9 billion in cash and cash equivalents against long-term debt of $301.6 million, a balance sheet that should give the company what it describes as plenty of runway to execute on its contracted pipeline.

Why does the order book matter?

The $11.9 billion figure represents contracted revenue already on the books, not speculative interest. That distinction matters because Fervo's recent record of hitting construction deadlines has, according to the company, driven new orders. As AI operators scour the market for firm, carbon-free electricity, geothermal has moved from niche to credible contender.

Fervo's flagship resource is Cape Station in Utah. The site could generate 8.4 gigawatts of electricity, a figure that doubles earlier expert estimates of roughly 4 gigawatts for the same resource. If fully built out, Cape Station alone would rank among the largest generation developments in the United States.

The company connected its first 100 megawatt power plant to the grid earlier this year. It is now on track to complete a 396 megawatt plant for Google by the end of 2028.

What did Tim Latimer say about the Google deal?

The Google project, now named Cape Google 400, will be grid connected, co-founder and CEO Tim Latimer said on the earnings call Monday. The structure gives Google flexibility: the company can draw power under the deal or sell it to other customers if it is not ready to consume it.

Latimer framed the arrangement as a scheduling decoupling that removes the usual chicken-and-egg problem between power plants and the facilities they supply. "We build Cape Station on our schedule. Google can develop a potential data center on their schedule, and neither party is waiting on the other," he said.

The quote underlines what may be geothermal's structural advantage over competing firm-power sources such as large nuclear reactors or natural gas plants: phased construction. Fervo does not have to finish an entire gigawatt-scale facility before delivering revenue-generating power.

How does a GeoBlock work?

Fervo's newer wells are deeper and hotter, and each is capable of generating 15 megawatts of gross power. The company links several wells together into what it calls a GeoBlock, a modular generating unit sized to match the demand of a typical data center hall.

The numbers show the concept maturing. The first completed GeoBlock generated 33 megawatts. Recent GeoBlocks have hit 50 megawatts. Because each block comes online as it is completed, hyperscalers can match power deliveries to their construction schedules rather than waiting years for a single massive plant.

That modularity is the commercial pitch. A hyperscaler can commission capacity hall by hall, in the same increments in which it builds data centers.

Can Fervo compete on price?

Much depends on cost. Hyperscalers are hungry for power today, but the company acknowledges they may not always be as price indiscriminate as they are during the current AI buildout.

The trend line so far favors Fervo. The first phase of Cape Station cost $7,000 per kilowatt of capacity, the company said. The Google project costs $5,500 per kilowatt, a drop of more than 21% in a single project cycle.

With subsidies included, that $5,500 per kilowatt is roughly equivalent to a new natural gas power plant running on fuel priced at $6 per million BTUs (British thermal units, the standard unit for comparing fuel energy content). Natural gas currently costs about half that level, so Fervo remains roughly twice as expensive as the incumbent fuel at today's prices.

The long-term target is $3,000 per kilowatt. At that level, Fervo could significantly undercut natural gas power even at current gas prices. The company calls the target ambitious, but the trajectory from $7,000 to $5,500 in a few years suggests the learning curve is real.

What happens next?

Fervo faces two tests. The first is execution: delivering the 396 megawatt Google project by the end of 2028 and continuing to hit deadlines, which the company credits for its growing order book. The second is economics: driving installed costs from $5,500 per kilowatt toward the $3,000 target before hyperscaler demand cools or loses its price tolerance.

The balance sheet supports the first test. The $1.9 billion cash position against $301.6 million in long-term debt gives the company room to fund construction while it burns $43.7 million per quarter at the operating level.

Google's early commitment suggests at least one hyperscaler has already run the numbers and decided Fervo can pass both. If the company keeps cutting costs at the pace it has demonstrated between Cape Station's first phase and Cape Google 400, geothermal could move from a premium option for AI power buyers to a direct competitor with natural gas on price.