Summary

  • Fervo’s signed agreement with Google covers 396MW for 15 years, with four successive 99MW sub-tranches targeted to begin commercial operation in the third quarter of 2028.
  • The approximately 600MW expansion remains an offer, not contracted capacity: Google must accept and the parties must negotiate a definitive agreement.
  • If Fervo does not make the required expansion offer, it may have to reimburse prior Google payments for deemed delivered energy under a formula whose amount is currently not estimable.

The most consequential sentence in Fervo Energy’s new Google contract is not the one that says “nearly 1GW.” It is the sentence that explains what happens if the second block is never properly offered.

On 26 August, Fervo’s Cape Generating Station 6 subsidiary entered into a power purchase agreement with Google Energy for a 396MW enhanced-geothermal project at Cape Station in Beaver County, Utah. The Form 8-K filed on 1 September sets out a commercial sequence that is more precise than the headline: four 99MW sub-tranches, target commercial-operation dates beginning in the third quarter of 2028, and a 15-year delivery term.

Those numbers reconcile. Four times 99MW equals the contracted 396MW. Multiplying nameplate capacity by the stated term gives 5,940 MW-years, but that is only a duration-weighted capacity measure. It is not megawatt-hours, expected generation, revenue or cash. The filing does not disclose the PPA price or the capacity factor.

The parent guarantees are another real receipt. Fervo guarantees the obligations of its project subsidiary; Alphabet guarantees Google’s obligations. The guarantees strengthen the counterparty chain. They do not remove construction, permitting, interconnection or operating risk.

The 600MW block has three gates

Fervo has agreed to offer Google an expansion of approximately 600MW. If it proceeds, the filing says total contracted enhanced-geothermal capacity would be not less than approximately 950MW, with a guaranteed commercial-operation date no later than June 2030. The filed press release calls the combined scale nearly 1GW.

That is not the same as saying nearly 1GW is under contract today. Fervo must first make the offer. Google must then accept it. Finally, both parties must negotiate a mutually acceptable definitive agreement. A required offer, a buyer option and a signed sale are three different states.

The asymmetry matters. The filing does not say Google owes a penalty if it declines. It says that if Fervo does not make the required offer, Fervo must reimburse Google for amounts previously paid to the project subsidiary for deemed delivered energy under the existing PPA. The amount would depend on two variables: the volume of deemed-delivery payments already made and the difference between the market price and the PPA price. Fervo says the result cannot currently be estimated.

This is not a fixed break fee, and no reimbursement is reported as due. It is a contingent economic tail attached to Fervo’s duty to place expansion terms in front of Google. The distinction gives the option more substance than a promotional ambition without turning it into booked capacity.

The contract is large relative to the June base

At 30 June, Fervo reported 658MW of binding PPAs and other power-sale arrangements, representing about US$7.2 billion of potential revenue backlog. The June-quarter 10-Q says that backlog assumes expected output, contracted prices including escalation or indexation, and full counterparty performance across the entire contract term.

The new 396MW equals 60.2% of that June contracted-MW base. If the new agreement is wholly additive to the reported 658MW, the arithmetic reaches 1,054MW. Fervo has not supplied an updated backlog value, however. Applying the old average dollar-per-MW relationship would be false precision: contract duration, output assumptions, price and escalation can differ.

The roughly 600MW option belongs outside both calculations. It is absent from the June backlog and remains outside signed capacity until acceptance and definitive documentation occur.

Offtake does not drill a well

Commercial visibility is only one conversion step. Fervo said Cape Station had 500MW under construction at 30 June: roughly 100MW in Phase I and 400MW in Phase II. It had received 82 of 179 government permits and approvals needed for Phase II commercial operation, leaving 97 in process in the 10-Q context.

Capital is equally tangible. Fervo expected Phase II to require about US$2.2 billion of cumulative capital expenditure through 2028 and said it intended to seek a significant portion as project-level debt. It spent US$399.3 million on capital expenditure in the first half of 2026 and projected another US$850 million to US$900 million for the rest of the year. Unrestricted cash stood at US$2.106 billion after the May IPO, while binding supplier commitments totalled US$488.3 million, mainly for the two Cape Station phases.

The US$421.4 million Project Granite facility should not be misread as the answer to Phase II. It finances Phase I. Fervo had US$212.3 million outstanding and US$157.8 million of undrawn borrowing capacity under that facility at June-end. Phase II financing remains a separate receipt.

Utah provides a pathway, not a completion certificate

Fervo intends to use flexible delivery routes authorised by Utah Senate Bill 132. The company says a closed private generation system could be configured with front-of-the-meter or behind-the-meter interconnection for data-centre load. Both the 8-K and the enrolled statute preserve conditions: engineering feasibility, agency approvals, system separation and approved arrangements where utility connections are involved.

The law creates a route through the regulatory map. It does not prove that Fervo’s route has been engineered, permitted or energised. Google also describes the Utah data centre as potential and subject to engineering, local approvals and commercial conditions.

The correct reading of the announcement is therefore neither “just an option” nor “one gigawatt secured.” A 396MW, 15-year PPA is signed and parent-backed. Four physical sub-tranches still need to arrive. A second block may expand the relationship, while a contingent reimbursement gives Fervo a reason to make the promised offer. Contract, option, sanction and operation belong on separate ledgers.

Sources