Next-Generation Geothermal Investment 2026–2030: What Fervo's Cape Station Proves — and What It Doesn't - Part I
By Green Fuel Journal Research & Intelligence Team
Last reviewed: October 2026 · Next review trigger: Fervo third-quarter 2026 results (date not yet announced)
GreenFuelJournal.com · Evidence cut-off 22 October 2026 · Primary reader: energy investors; secondary readers: utility and hyperscale data-centre executives.
Legal Disclaimer
This report is for information only and is not investment, legal or tax advice. It does not recommend buying, selling or holding any security, including shares of Fervo Energy (Nasdaq: FRVO). Information is current to 22 October 2026. Company figures are company-reported and have not been independently audited unless stated. Statements by Morpheus Research (a short seller), Hagens Berman (a plaintiffs' law firm) and GeoExpro (commentary) are unverified third-party allegations or opinions, attributed to their authors. Scenarios, targets and the illustrative calculation in Section 7.2 are not forecasts. AI tools supported the research and drafting, and human editors reviewed the report; verify critical figures against the cited sources.
Position disclosure: Green Fuel Journal, its authors and SEKASON RESEARCH LTD hold no position in Fervo Energy or competing securities as of 22 October 2026.
Read the full GreenFuelJournal.com disclaimers.
Executive Summary
Fervo geothermal brought GeoBlock 1 at Cape Station in Utah (33 MW net) to commercial operation on 30 September 2026, but the unit has about three weeks of operating history and no independent party has confirmed its cost or reservoir performance. The Cape Station cost figures, about US$7,000/kW for Phase I, a US$5,500/kW Phase II target and US$3,000/kW long term, all originate with Fervo, against Lazard's US$5,135–6,635/kW for conventional geothermal. Capital has moved ahead of that evidence: Fervo Energy raised about US$2.2bn gross in its May 2026 Nasdaq listing and had drawn US$212.3m of a US$421.4m Phase I facility by 30 June 2026.
Offtake is large but not yet reconciled. Fervo reported 658 MW of binding power purchase agreements at its IPO, worth about US$7.2bn of revenue backlog, and its chief financial officer said in October 2026 that the portfolio holds “over 1 gigawatt” of binding agreements; the two figures have not been reconciled. Google added a 396 MW agreement on 1 September 2026. About 290 MW of reported transmission rights sit against 384 MW of contracted Phase II capacity (single secondary source).
Two questions decide whether Cape Station supports an investable asset class. The first is reservoir life: Fervo's Project Red pilot reports about 20% thermal recovery and a 2.5°F temperature dip after roughly 500 days, while its resource estimates assume 30.5% and one critic cites 5–15% (opinion, unverified). The second is delivery of Phase II (400 MW) and the Google contract, both due in 2028. Treat 2026–2027 as the evidence-building period and 2028–2030 as the replication test.
1. Executive Intelligence Synthesis
What did Cape Station's commercial operation prove?
Cape Station proved that a greenfield enhanced geothermal unit can be built and declared commercial under a power purchase agreement: GeoBlock 1 (33 MW net) reached commercial operation on 30 September 2026, about 23 months after construction began. It did not prove reservoir life, because the unit has about three weeks of history. It did not prove independent cost levels, because the Phase I cost figure originates with Fervo. It did not prove replication, because Phase II sits beside Phase I.
1.1 The five Executive Signals
Five signals summarise the evidence: operating proof is real but short, capital is not the near-term constraint, cost evidence is company-sourced, offtake is unreconciled, and reservoir life and power delivery remain open. Each states the finding, why it matters to an energy investor, and a specific test to apply.
Signal 1 · Operating proof is real but short.
FINDING: GeoBlock 1 delivered 33 MW net and was declared commercial on 30 September 2026, with grid synchronisation on 24 September 2026 (company-reported, grade F).
SO WHAT: A delivered unit shows that construction, connection and contracting can work together; three weeks of output say nothing about how the reservoir behaves over a 15-year contract.
NOW WHAT: Require 12 months of Cape Station wellhead temperature and net megawatt data before underwriting any later phase.
Signal 2 · Capital is not the near-term constraint.
FINDING: Fervo held US$2,106.4m of cash at 30 June 2026 against second-half 2026 capital expenditure guidance of US$850.0–900.0m (grade F).
SO WHAT: Funding risk through 2026 is low; the open funding question is Phase II debt, which the filings read for this report do not disclose.
NOW WHAT: Ask what share of Phase II is covered by committed debt, and on what recourse terms, before treating 2028 delivery as funded.
Signal 3 · Cost evidence is company-sourced.
FINDING: Phase I cost of about US$7,000/kW sits above Lazard's conventional geothermal capital range of US$5,135–6,635/kW; the US$5,500/kW and US$3,000/kW figures are Fervo targets (grade C, [SINGLE-SOURCE]).
SO WHAT: The investment case rests on projected cost declines, not on present cost parity with other geothermal.
NOW WHAT: Model returns at US$7,000/kW first, and treat any lower figure as an upside case until an independent party confirms it.
Signal 4 · Offtake is strong but unreconciled and concentrated.
FINDING: Fervo reported 658 MW of binding PPAs at its IPO and, in October 2026, “over 1 gigawatt” of binding PPAs (CFO David Ulrey, Fervo release of 1 October 2026); Google's 3 GW framework is non-binding beyond the 396 MW agreement, according to Deseret News on 14 October 2026 (grade F).
SO WHAT: The two binding figures have not been reconciled in a primary source, so the revenue backlog cannot yet be tied to named megawatts by site.
NOW WHAT: Request a contract schedule by counterparty, megawatt, price and delivery date, and count only binding volumes.
Signal 5 · Reservoir life and power delivery remain open.
FINDING: Fervo's Project Red pilot reports a thermal recovery factor of about 20% against a 30.5% resource assumption, and Eavor's Geretsried site was reported at 0.5–2.0 MWe gross against an 8.2 MWe target (grades C and D).
SO WHAT: Announced capacity and delivered output can diverge at first-of-a-kind sites, and the reservoir assumption governs the size of later phases.
NOW WHAT: Make an independent reservoir review and a firm interconnection a condition of any Phase II funding decision.
Table 1. Five Executive Signals and their evidence grades · as of 22 October 2026
Signal | Anchor evidence | Grade |
1. Operating proof | 33 MW net, commercial 30 September 2026 | F |
2. Capital | US$2,106.4m cash vs US$850.0–900.0m guidance | F |
3. Cost | US$7,000/kW, Fervo-origin; no independent check | C |
4. Offtake | 658 MW at IPO vs “over 1 gigawatt”; not reconciled | F D |
5. Reservoir and delivery | 20% vs 30.5% recovery; 290 MW rights vs 384 MW contracted | C D |
Source: Fervo Energy Q2 2026 earnings release (12 August 2026), Form 10-Q (30 June 2026), S-1/A and 424B4 prospectus; Lazard LCOE+ v19.0 (July 2026); ThinkGeoEnergy (16 October 2026). Grades defined below.
Methodology note. Grades follow the key in Section 1.2 and describe the quality of the evidence behind each row, not the quality of the project. Each row carries the grade of the evidence cited for it, and no overall score or weighting is calculated.
1.2 The evidence-grade key
No Executive Signal reaches the top evidence grade, and the strongest any of them reaches is a company filing or release. Every claim in this report carries one of five grades, which describe the quality of the evidence, not the quality of the project.
FINDING: None of the five Executive Signals rests on an independently verified source (grade V); the strongest grade in Table 1 is F, a company filing or release.
SO WHAT: Every headline fact about Cape Station is currently reported by Fervo or by an interested third party, so a reader cannot yet cross-check it against a regulator, laboratory or auditor.
NOW WHAT: Re-rate a signal only when a grade V source appears, and ask Fervo which independent parties are reviewing its data.
Table 2. Evidence grades used throughout this report
Grade | Meaning | What qualifies |
V | Independently verified | A regulator, laboratory or audited third party confirms the claim |
F | Primary filing or release | Company-reported in an SEC filing or official release |
C | Company claim, uncorroborated | The company states it; no second origin found |
D | Contested or allegation | Disputed, or raised only by an interested third party |
N | Not available | No source found in the sources reviewed |
Cost figures also carry [2-SOURCE], [SINGLE-SOURCE] or [CONFLICT] labels, explained in Section 7.2. Figures calculated by GFJ are marked "GFJ arithmetic" or "GFJ inference".
1.3 What Cape Station has and has not settled
FINDING: Construction and commissioning risk for one 33 MW block is now documented by the company (grade F); no independent party has yet confirmed resource, cost or grid-delivery risk (grade N).
SO WHAT: Only one of the six bankability dimensions has moved from forecast to observed result, and it is the dimension with the shortest track record.
NOW WHAT: Score each later announcement against the Section 7.3 scorecard before changing an investment view.
Cape Station has narrowed one uncertainty: whether a developer can drill, stimulate, connect and sell enhanced geothermal power on a contract schedule. It has not narrowed the three that carry most of the value, which are how long the reservoir holds temperature, what the next block really costs, and whether the power reaches buyers on time.
2. Macro Context & Strategic Drivers
Why is next-generation geothermal attracting investment now?
Three pressures coincide. The International Energy Agency (IEA) projects data-centre electricity use rising from 415 TWh in 2024 to about 945 TWh in 2030. The IEA also reports sector financing of about US$2.2bn in 2025, up 80%. Whether geothermal kept its full United States tax-credit eligibility under the law enacted on 4 July 2025 is unconfirmed, because the geothermal-specific text was not located. The IEA still describes the technology as at an early stage of development.
2.1 Demand: data centres want firm power
FINDING: The IEA's Energy and AI base case puts data-centre electricity use at 415 TWh in 2024, about 945 TWh in 2030 and about 1,200 TWh in 2035 (grade V as an IEA projection, not an observation).
SO WHAT: Demand is rising faster than firm clean supply, yet the IEA report mentions geothermal once and gives no geothermal figure, so any claim that geothermal will supply a stated share of data-centre load comes from companies.
NOW WHAT: Treat buyer demand as established and geothermal's share of it as unproven; ask each developer for signed megawatts, not market-share estimates.
The buyer evidence is concrete. Google signed a 396 MW agreement with Fervo for Cape Station on 1 September 2026, with power due online in 2028 and an option on about 600 MW more by June 2030. The IEA notes that recent firm clean power contracts have in some cases reached about US$130/MWh, against US$30–60/MWh earlier. That price signal explains why a developer with costs above the conventional range can still sign contracts.
For GFJ's wider coverage of how artificial intelligence is reshaping power demand, see Energy AI Agents 2035.
2.2 Capital: where the money is going
FINDING: The IEA reported about US$2.2bn of next-generation geothermal financing in 2025, up 80%, and Wood Mackenzie reported US$1.5bn in the third quarter of 2026 and US$5.5bn year to date (grades F and C; the Wood Mackenzie figures come from a paywalled summary).
SO WHAT: Money is flowing into the sector, but the two series cover different periods and cannot be added or compared year on year.
NOW WHAT: Use each series only for the period it describes, and ask the data provider which transactions it counts.
Table 3. Capital flows into next-generation geothermal, by source and date
Source or company | Figure | Period or date | Note |
IEA | About US$2.2bn, up 80% | Calendar 2025 | Sector financing |
Wood Mackenzie | US$1.5bn | Q3 2026 | Paywalled summary |
Wood Mackenzie | US$5.5bn | Year to date, 15 October 2026 | Wood Mackenzie separately reports a US$315m Quaise and Mazama raise |
Fervo Energy IPO | About US$2.2bn gross (about US$1.89bn base deal) | May 2026 | At US$27 a share |
Fervo Energy Series E | US$462m | About 10 December 2025 | Led by B Capital |
Sage Geosystems Series B | Over US$97m | 21 January 2026 | Co-led by Ormat and Carbon Direct Capital |
Quaise Energy Series B | US$134m | July 2026 | A second outlet reports US$144m; not reconciled |
Eavor Technologies Series B | C$182m | 25 October 2023 | Includes C$90m from Canada Growth Fund |
Source: IEA commentary (23 January 2026); Wood Mackenzie (15 October 2026); Fortune (14 May 2026); Business Wire (10 December 2025, 21 January 2026); EnergyConnects (July 2026); BetaKit (25 October 2023). Wood Mackenzie's earlier US$2.2bn cumulative figure (2019 to mid-2025, published 27 August 2025) is out of date and is not a current total.
The GFJ inference is narrow: Fervo's IPO alone raised a sum of the same order as the IEA's whole-sector 2025 figure; the two figures cover different scopes and periods. Whether the IEA counts public listings is not stated. For how corporate buyers and investors are pricing energy-technology deals more broadly, see our New Energy M&A Playbook 2026–2027.
2.3 Policy: a United States advantage, patchy elsewhere
FINDING: The 45Y and 48E clean-electricity credits carry foreign-entity-of-concern material-assistance ratios rising from 40% in 2026 to 60% from 2030 under the law enacted on 4 July 2025; whether geothermal keeps full eligibility is unconfirmed (grade C: a law-firm alert gives the general schedule, and the geothermal-specific text is unconfirmed).
SO WHAT: The credit gives United States projects a financing edge that EU and Asian projects do not currently match, but its application to turbines and drilling inputs needs checking project by project.
NOW WHAT: Obtain tax counsel's reading of the phase-out year and the material-assistance test for each project's construction start date.
Permitting is moving in the same direction. The Bureau of Land Management (BLM) told a House subcommittee on 16 December 2025 that it had adopted two geothermal categorical exclusions, developed a third and was using emergency permitting procedures; it manages 569 geothermal leases covering about 1.2m acres. Eight geothermal bills cleared the House Natural Resources Committee on 5 March 2026, and none had been enacted at the time of that source. The Department of Energy announced US$99M for 21 geothermal projects on 21 September 2026, five led by Utah-based organisations.
Outside the United States the instruments are weaker. The EU's promised Geothermal Action Plan was still unpublished in mid-August 2026 and was reported to have dropped off the Commission's agenda. Germany issued its Geothermie-Beschleunigungsgesetz (GeoBG) on 22 December 2025, giving geothermal overriding public interest until greenhouse-gas neutrality is reached, which the text places in 2045.
The IEA analysts Rebecca Schulz, Martina Lyons, Deniz Ugur, Simon Bennett and Courtney Turich wrote in an IEA commentary on 23 January 2026 that
"next-generation geothermal technology remains at an early stage of development."
hat sentence frames the whole policy picture: support is arriving before the evidence is complete.
3. India Implications
Can India replicate Fervo's Cape Station?
Not on current evidence. India has no commercial-scale geothermal power; its pilots are kW-scale: a 20 kW closed-loop plant announced in Telangana in 2024 and 50–90 kW at Gandhinagar. The largest planned project is a 1 MWe pilot at Puga in Ladakh, run by the ONGC Energy Centre, where two wells were commissioned in July 2026. The national policy reported on 17 September 2025 sets no numeric capacity target, and most of the resource is low to medium enthalpy.
FINDING: India's geothermal resource is put at 10,600 MW, a theoretical figure from the Geothermal Atlas of India (2022), and ThinkGeoEnergy (17 September 2025) reports that most of it lies at 100–180°C (grade F, Press Information Bureau document of 4 September 2026).
SO WHAT: Cape Station-style hard-rock enhanced geothermal is not an obvious near-term fit for a resource of that temperature, and no US cost or performance figure in this report transfers to Indian geology or market conditions.
NOW WHAT: Treat India as a monitoring case for 2027–2028 and look for a numeric capacity target, a funded Puga investment figure and a named commercial developer before allocating capital.
The government evidence is clear about scale. A Press Information Bureau (PIB) document dated 4 September 2026 records no commercial-scale geothermal power in India. The Ministry of New and Renewable Energy's National Policy on Geothermal Energy, reported on 17 September 2025, offers exploration permits of 3 years (extendable by 2), development allocations of up to 30 years, state single-window agencies, a national data repository, 100% foreign direct investment in renewables and conversion of abandoned oil and gas wells. The press reports reviewed give no numeric capacity target, royalty or fee, and the official policy text was not located.
The Puga pilot is operated by the ONGC Energy Centre with the Ladakh administration, and its memorandum of understanding was extended by five years (reported 23 May 2026). No investment figure for Puga has been published. Reports of the reservoir temperature at Puga conflict, so this report prints none. Baker Hughes and Indian officials were reported on 21 July 2026 to be in talks on geothermal cooperation; no agreement has been announced. We found no link between any Indian entity and Fervo, Eavor or Quaise.
Table 4. India: six investor questions, available evidence and grade · as of 22 October 2026
Investor question | What evidence exists | Grade |
1. Where could EGS work commercially? | 10,600 MW theoretical resource; mostly 100–180°C; no EGS site assessment found | F N |
2. What would the electricity cost? | No cost per MW or per MWh for India found | N |
3. Can oil and gas capability lower costs? | ONGC Energy Centre runs Puga; policy allows conversion of abandoned wells; Baker Hughes talks exploratory only | F N |
4. What approvals and incentives apply? | Permits, allocations and single window as press-reported; official text not located | N |
5. Who could finance first projects? | No Puga investment figure; no named commercial financier found | N |
6. Is there a credible first customer market? | No procurement, tariff or offtake evidence found | N |
Source: PIB document (4 September 2026); ThinkGeoEnergy and Renewable Watch (17 September 2025); Business Standard/PTI (23 May 2026); Construction World (21 July 2026); Drishti IAS (22 July 2026). Questions are GFJ editorial questions for this report.
Methodology note. Each question is graded on the best evidence found for it, using the key in Section 1.2; grade N means no evidence was found. The grades are descriptive, are not combined into a score, and carry no weighting.
A Cape-style transfer would need four things in sequence: a site-level resource assessment at depth, drilling capacity that Indian oil and gas operators could supply, a partner with enhanced geothermal field experience, and a buyer willing to sign a long-term contract. Only the second has any visible foundation today.
4. Operational & Technical Deep-Dive
What is the biggest technical risk to enhanced geothermal returns?
Thermal decline is the biggest technical risk: how fast the reservoir cools, and how much of its heat can be recovered over a 15-year contract. Fervo's Project Red pilot reports a recovery factor of about 20%, while Fervo's resource estimates assume 30.5%, and one critic cites 5–15% (opinion, unverified). No independent reservoir review of Cape Station exists in the sources reviewed, and GeoBlock 1 has about three weeks of operating history.
4.1 What Cape Station demonstrates technically
FINDING: GeoBlock 1 reached grid synchronisation on 24 September 2026 and commercial operation on 30 September 2026, about 23 months after construction began, with an 18-month target for later blocks (company-reported, grade F).
SO WHAT: Delivery speed is the one technical claim already tested in the field, and a faster second and third block would be the first evidence of repeatable construction.
NOW WHAT: Track the dates of the remaining two Phase I GeoBlocks against the 1 January 2027 target and compare each build time with the 23 months.
Tim Latimer, chief executive and co-founder of Fervo, stated in Fervo's commercial-operation release on 1 October 2026:
"No team has ever built a project like this anywhere in the world, and we did it ahead of schedule."
It is a company statement, and Wood Mackenzie separately describes Cape Station as the largest operational enhanced geothermal plant "by a significant margin" (paywalled summary, 15 October 2026).
Dawn Owens, SVP, Head of Development and Commercial Markets at Fervo, stated in the same release on 1 October 2026:
"We are on track to bring the remaining Phase I GeoBlocks online by January 1, 2027."
Phase I totals about 100 MW, built as three GeoBlocks of 33 MW each. Phase II is a further 400 MW in eight blocks of 50 MW, expected in 2028. Drilling data are better than the construction record. Well Sawtooth 7, the Phase II design well, reached a total depth of 19,448 ft with a 7,500 ft lateral in 21 days from spud to total depth, in a resource of about 460°F (Fervo Q2 2026 release, 12 August 2026; Heatmap, 9 July 2026). Heatmap reports that this is nearly 35% deeper and about 50% longer in lateral than the Phase I average, which was about 14,000 ft deep with a 5,000 ft lateral (424B4 prospectus). Fervo reports that drilling time fell about 75% and per-foot drilling cost fell about 70% between 2022 and 2025 (company-reported, grade C).
One caution applies to every drilling-cost figure. The only dated per-well cost is from 12 February 2024, when Fervo reported a fall from US$9.4m to US$4.8m across its first four Cape horizontal wells. Current per-well cost for the Phase II design is not public. The US Department of Energy's Liftoff report cites a fall from US$13m to under US$5m per well at the Utah FORGE research site, which is a government-funded test site and not Cape ([2-SOURCE]).
4.2 The unresolved issue: thermal decline and recovery factor
FINDING: Fervo reports that Project Red recovered about 20% of its estimated initial heat in place after 614 days, while Fervo's own resource estimates assume 30.5% (company-reported, grade C; Fervo, 13 April 2026, and the S-1/424B4 prospectus).
SO WHAT: The recovery factor sets how much power a given block of rock can supply, so a gap between 20% and 30.5%, or a lower critic range, moves the size of every later phase and not the contracted megawatts of Phases I and II.
NOW WHAT: Ask Fervo to publish Cape Station wellhead temperature and net megawatts monthly, and ask any lender to commission a reservoir review that does not rely on Fervo's model.
What Fervo reports. Project Red is a small pilot in Nevada. Fervo's data, dated 13 April 2026 and modified 22 May 2026, show the project on production for more than 614 days. Fervo describes it as a learning platform, not a production-optimised system, and says gross power is modelled from an organic Rankine cycle model and weather data rather than metered.
Table 5. Project Red reported metrics (company data) · as of 13 April 2026
Metric | Reported value | Note |
Days on production | More than 614 | Page title says "two years"; body says 614 days |
Average gross output | 2.1 MW | Modelled, not metered |
Average net output | About 1.4 MW | |
Average production temperature | 347°F | |
Flow | About 36 kg/s | |
Uptime | 98.4% | Excludes surface and grid outages |
Temperature decline | About 2.5°F near day 500 | Fervo says this matches its model |
Water recapture | About 70% | |
Thermal recovery factor | About 20% | Of estimated initial heat in place |
Source: Fervo Energy article on two years of Project Red production data (13 April 2026, modified 22 May 2026). All values company-reported, grade C. No independent confirmation was found.
The prospectus states that Project Red has not experienced premature thermal decline, and it lists well and plant underperformance as a risk. It also says that resource estimates depend on assumed recovery factors that are subject to change with additional data (424B4 prospectus, May 2026; only the first part of the document was read for this report).
The disputed assumption (opinion). Elliot Yearsley, writing in the trade magazine GeoExpro on 3 June 2026 (opinion, information as of 19 May 2026), argues that the S-1 relies on a heat-initially-in-place report from DeGolyer and MacNaughton and not on a reserves report. The S-1 itself says its resource figures should not be viewed as a measure of reserves under SEC guidelines. Yearsley contrasts Fervo's 30.5% recovery assumption with a range of 5–15% that he attributes to Libbey and Murphy (2026) for low-productivity-index reservoirs like Cape, which is two to six times lower. We did not open that paper and cannot confirm his reading. He also writes that a reservoir model co-authored by Fervo uses assumptions rather than history-matched production, that no third-party review of Cape well testing appears to exist, and that only 24 hours of data from a 30-day flow test were shown to support a claimed 8–10 MW from the first Cape production well. In a second piece on 27 July 2026, GeoExpro says Project Red started experiencing thermal decline and needed an external source of injection water, without giving a rate. No Fervo response to either article was found.
The allegations. Morpheus Research, a short seller that holds a financial interest in a falling share price, published a report on 6 October 2026. It alleges a decline rate of about 24°F a year at Project Red, a 35% water loss in June 2026 and about 193m gallons leaked from December 2023 to July 2026. It also alleges that a March 2024 injection test showed temperature falling about 90°F while water loss rose from 21% to 57%, and says Fervo's April 2026 article calls the loss temporary. A plaintiffs' law firm, Hagens Berman, repeats a water loss of about 30% at Project Red in a release of 13 October 2026; that magnitude is consistent with Fervo's own roughly 70% recapture, and the dispute is about what it means at commercial scale. Neither is verified, both authors have a financial or legal interest, and no Fervo response to Morpheus was found. Grade: D.
The decline figures cannot be reconciled. Morpheus cites about 24°F a year, and a former employee, as relayed by Morpheus, calls 1–1.25°F a year normal for hydrothermal plants. Fervo's own 2.5°F dip over roughly 500 days is the only decline the company acknowledges. None of these figures is verified.
GFJ inference ·
FINDING: Fervo's Project Red figure of about 20% sits between its own 30.5% assumption and the 5–15% critic range, so the number is unsettled in both directions.
SO WHAT: If 5–15% applied in place of 30.5%, the resource for a given heat-in-place figure would be roughly 50–85% lower (GFJ arithmetic); that would not change the contracted Phase I and II megawatts, but it would bear on the 4.3 GW build-out potential, later phases and re-drilling capital costs.
NOW WHAT: Size any investment in Phase III or later against the 5–15% case, and treat the 30.5% case as the upside.
Source: Fervo Energy (13 April 2026); Fervo S-1/424B4 (2026); GeoExpro (3 June 2026).
What would resolve the question is specific. Twelve to 24 months of Cape Station production temperature and net megawatts from 30 September 2026 would be the first independent-grade evidence if Fervo discloses wellhead data. A neutral reservoir audit and Fervo's response to the Morpheus and GeoExpro claims would complete it. The register below shows what is public today.
Table 6. Evidence register: Cape Station thermal and cost data · as of 22 October 2026
Item | Status | Grade |
Project Red long-run summary | Public; company-reported | C |
Cape wellhead temperature series | Not public | N |
Per-well flow rates for Phase I wells | Not public | N |
Current per-well cost, Phase II design | Not public; latest dated figure is from February 2024 | N |
Independent reservoir review | None found | N |
Third-party review of Cape well tests | None found | N |
Microseismic catalogue | Exists on the DOE Geothermal Data Repository; not opened for this report | N |
Fervo response to Morpheus and GeoExpro | None found | N |
Source: GFJ review of Fervo filings and releases, DOE Liftoff fact sheet (October 2024), GeoExpro and Morpheus Research.
Methodology note. Each row is graded on the type of source behind it (Section 1.2): company filing or release, secondary report, third-party allegation or opinion. The register ranks nothing and applies no scoring formula; where sources conflict, both figures are shown.
4.3 Technology pathways compared
FINDING: Only the enhanced geothermal approach has a commissioned commercial unit; closed-loop and superhot-rock designs remain below target or pre-commercial (grades F and D).
SO WHAT: The pathways are at different maturity stages, so their financing risk is not comparable, and no data in the sources reviewed name a winner.
NOW WHAT: Compare developers on delivered megawatts and disclosed costs per kilowatt, not on announced capacity.
Enhanced geothermal systems (EGS), used by Fervo, create flow paths in hot rock by drilling and stimulation; Sage Geosystems uses a related pressure-based design it calls Pressure Geothermal; Fervo has a 33 MW net unit in commercial operation, and Sage has an agreement with Meta for up to 150 MW with a first phase slated for 2027 (a 2024 article). Closed-loop systems, used by Eavor and XGS Energy, circulate fluid in sealed wells; Eavor's Geretsried site has been reported well below target, and XGS has announced a 150 MW New Mexico project with no disclosed timeline or cost. Superhot-rock drilling, pursued by Quaise Energy, targets much hotter rock; Quaise plans 250 MW in Oregon, with a flow test planned for the end of 2026 and a 5,000 m target by the end of 2027.
5. Named Company Case Studies
How do Fervo, Eavor, Sage and XGS compare?
Only Fervo has a commissioned commercial unit: GeoBlock 1 (33 MW net) was declared commercial on 30 September 2026. Eavor's Geretsried site was reported on 16 October 2026 at 0.5–2.0 MWe gross against an 8.2 MWe target. Sage Geosystems and XGS Energy have agreements and plans, but no commissioned commercial unit appears in the sources reviewed. Pertamina Geothermal Energy and Kaishan operate conventional plants and serve as baselines, not as evidence for enhanced geothermal.
5.1 Fervo Energy: a funded delivery record, an unfinished evidence base
FINDING: Fervo listed on Nasdaq in May 2026 at US$27 a share, raising about US$2.2bn gross including the over-allotment, and had drawn US$212.3m of US$421.4m in Project Granite commitments by 30 June 2026 (grade F, Form 10-Q).
SO WHAT: Public equity and project-level lenders have both accepted Fervo's contracts and construction record, which removes funding as the near-term constraint.
NOW WHAT: Separate what lenders accepted for one project from what any other developer could raise, and ask for the Project Granite loan terms before using it as a benchmark.
Project Granite is described in Fervo's own financing release of 19 March 2026 as non-recourse project financing, meaning lenders look to the project's cash flows and not to Fervo's wider balance sheet. It closed in the first quarter of 2026, and ESG Dive reported the lenders as Barclays and HSBC. The offtake side is substantial: Southern California Edison signed two 15-year agreements for 320 MW on 25 June 2024, Google holds 115 MW in Nevada through NV Energy and 396 MW from Cape Station, and Clean Power Alliance and Desert Community Energy also hold contracts.
Delivery has matched the schedule. Grid synchronisation and commercial operation both came earlier than the Q2 2026 release's guidance of test power in the fourth quarter. A well blowout on 25 May 2026, reported by Utility Dive, did not change that: Fervo said there were no injuries, no environmental damage and no material effect on cost or schedule.
Three conflicts remain open.
First, Fervo's release does not name the buyer of GeoBlock 1; IEEE Spectrum reports that Fervo will supply Shell Energy once commercial operations begin, without tying the contract to the first unit, but no primary document confirms it ([SINGLE-SOURCE]).
Second, Fervo reports 658 MW of binding agreements at its IPO and “over 1 gigawatt” in October 2026, and no primary source reconciles the two.
Third, balance-sheet debt of about US$228m at 30 June 2026 (US$11.0m current and US$217.4m long term) differs from the US$212.3m drawn on the facility, and the difference is unexplained. The two IPO figures do not conflict: about US$1.89bn is the base deal of 70 million shares at US$27 (GFJ arithmetic), and about US$2.2bn is the 80.5 million shares sold including the over-allotment (Form 10-Q).
The record supports one reading: a developer can be financed, contracted and built on time. It does not support three others, which are independent confirmation of cost, reservoir life beyond three weeks at Cape, and repeatability outside Utah and Nevada.
5.2 Eavor Technologies: what an announced output can hide
FINDING: ThinkGeoEnergy reported first grid electricity at Geretsried, Bavaria, on 4 December 2025, then reported on 16 October 2026 that Eavor had withdrawn its two rigs and that the site had delivered 0.5–2.0 MWe gross, against an 8.2 MWe target first reported on 4 December 2025 (trade press, grade D).
SO WHAT: The two reports from the same outlet differ sharply, and Eavor's own metered output was not found, so the output of the closest closed-loop peer cannot be confirmed.
NOW WHAT: Ask any developer for metered net output over a stated period and treat first-power announcements as milestones, not production.
Public financing was substantial. The European Investment Bank lent about EUR 45m and the EU Innovation Fund granted EUR 91.6m, with a company-stated total investment of EUR 350m (EIB press release 2024-168, 2 May 2024; [SINGLE-SOURCE]). The design called for two wells of 4,500–5,000 m with 12 lateral pairs per well. Eavor's Series B reached C$182m in October 2023. A forum repost of a POWER Magazine item adds an 8 MWe and 64 MWth figure that we could not verify. For investors, Geretsried shows that the gap between announced and delivered output can open at a first-of-a-kind site, even with public funding behind it. It does not show that closed-loop technology cannot reach its target, because no operator statement after December 2025 was found.
5.3 Pertamina Geothermal Energy: the conventional baseline
FINDING: Pertamina Geothermal Energy (PGE) operates 727 MW and reported first-half 2026 revenue of US$235.03m, up 14.7%, and net profit of US$79.6m, up 15.48% (revenue and profit: InvestorTrust, 4 August 2026; capacity: KB Valbury, 6 August 2026; grade F as relayed).
SO WHAT: A proven geothermal operator earns recurring revenue at scale, which is the reference point for reading Fervo's US$113k of second-quarter 2026 revenue.
NOW WHAT: Use PGE's margins and capital plans as a reference case for any conventional geothermal assumption in a model.
PGE's president director, Ahmad Yani, told the IIGCE conference on 20 August 2026 that the company targets 1 GW by 2028 and 1.8 GW by 2034 (Indonesia Business Post, 21 August 2026). One broker, KB Valbury, lists a pipeline including Hululais 1 and 2 (2 × 55 MW, 2028 forecast) and Kotamobagu (64 MW, 2030 forecast), and forecasts capital spending of US$102m (2025 actual), US$130m (2026), US$482m (2027) and US$585m (2028). These are broker forecasts, not company guidance, and the broker's 1.85 GW by 2033 differs from the company's 1.8 GW by 2034. PGE has no enhanced geothermal project, so its case shows what conventional geothermal economics look like and says nothing about whether enhanced systems will match them.
5.4 Kaishan and OrPower 22: a timeline benchmark from Kenya
FINDING: Kaishan's subsidiary OrPower 22 brought the 35 MW Menengai III plant to commercial operation on 10 March 2026, taking Kenya past 1 GW installed (ThinkGeoEnergy, 9 April 2026, grade C).
SO WHAT: It gives a recent date-stamped example of a conventional geothermal plant delivered under a power purchase agreement with Kenya Power and Lighting Company, but no investment figure was found, so no cost per kilowatt can be derived.
NOW WHAT: Use the case for schedule and capacity comparison only, and request the investment figure before comparing it with Cape Station costs.
Testing finished on 5 March 2026. Estimated output is about 291.27 GWh a year and estimated revenue about US$15m a year ([SINGLE-SOURCE]).
GFJ arithmetic: 291.27 GWh from 35 MW implies a capacity factor near 95%, an estimate and not reported output. Kenya's average geothermal tariff is about US$0.069/kWh (Business Daily Africa, 17 September 2025).
5.5 Peer comparison
FINDING: None of the four developers other than Fervo has disclosed a cost per kilowatt or a metered operating output in the material reviewed (grade N).
SO WHAT: Developers cannot be ranked on cost or performance today, only on stage, financing and contracted volume.
NOW WHAT: Ask each developer for metered net output and a cost per kilowatt on a stated basis before comparing it with Fervo.
The table sets the five next-generation developers side by side. "Not disclosed" means the research found no figure; it does not mean none exists.
Table 7. Next-generation geothermal peers: stage, financing, offtake and evidence of operation
Developer | Design | Stage | Financing | Offtake | Evidence of operation | As of |
Fervo Energy | EGS | Commercial unit | IPO about US$2.2bn; Series E US$462m; Project Granite US$421.4m | 658 MW binding at IPO; Google 396 MW | 33 MW net, 30 September 2026 | 22 Oct 2026 |
Eavor | Closed loop | Below target | EIB about EUR 45m; Innovation Fund EUR 91.6m; Series B C$182m | Firm heat offtake under negotiation | 0.5–2.0 MWe gross vs 8.2 MWe target | 16 Oct 2026 |
Sage Geosystems | Pressure Geothermal | Pre-commercial | Series B over US$97m | Meta, up to 150 MW; first phase slated 2027 | Not disclosed; pilot with Ormat in Nevada reported | Jan 2026 (financing); Aug 2024 (offtake) |
XGS Energy | Closed loop | Pre-commercial | Not disclosed | 150 MW New Mexico for Meta; no timeline | Not disclosed | 22 May 2026 |
Quaise Energy | Superhot rock | Pre-commercial | Series B US$134m (one outlet: US$144m) | 250 MW Oregon plan; hyperscaler committed to first 50 MW | Oregon flow test planned end-2026 | Jul 2026 |
Source: Fervo Form 10-Q and releases; ThinkGeoEnergy (16 October 2026); EIB press release 2024-168; Business Wire (21 January 2026); Canary Media (26 August 2024); Data Center Dynamics (22 May 2026); EnergyConnects (July 2026); Ormat Q2 2026 results (5 August 2026). Dates differ by column.
6. Friction, Risk & Systemic Bottlenecks
What could stop Cape Station delivering contracted power?
Three things could: transmission capacity, reservoir performance and construction timing. About 290 MW of interconnection rights were reported against 384 MW of contracted Phase II capacity (MLQ.ai, 5 May 2026, single secondary source). Reservoir performance at Cape has about three weeks of data. The remaining Phase I blocks are due by 1 January 2027. Third parties have raised further concerns on each point, but none is verified, and no Fervo response was found.
Every item below is labelled by origin: [COMPANY] for Fervo's own statements, [FILING] for SEC documents, [PRESS] for news and analyst reports, and [ALLEGATION] or [OPINION] for material from Morpheus Research (a short seller), Hagens Berman (a plaintiffs' law firm) or GeoExpro (commentary). Allegations are reported so that readers can see what has been said, not because they are established.
6.1 Delivery: transmission and interconnection
FINDING: Cape Phase II holds about 290 MW of interconnection rights against 384 MW of contracted capacity, a 94 MW gap (MLQ.ai, 5 May 2026; gap is GFJ arithmetic; grade C, single secondary source).
SO WHAT: Contracts are only worth the megawatts that can reach a buyer, so an unresolved transmission path puts the 2028 delivery dates at risk before any reservoir question arises.
NOW WHAT: Require a firm interconnection agreement, with route, owner and in-service date, as a condition of funding Phase II.
Julien Dumoulin-Smith, an equity analyst at Jefferies, wrote in a note reported on 5 June 2026 that transmission constraints could slow Fervo's roll-out because most sites are in rural areas with limited infrastructure. The 424B4 prospectus lists dependence on third-party transmission systems, including potential curtailments or recalls, and limited transformer availability, as risk factors. Fervo is reported to have told the California Public Utilities Commission in March 2026 that urgent action is needed to deliver power through new transmission; we found that filing but did not read it.
Morpheus Research alleges that Fervo told California's grid operator in April 2026 that there was no viable pathway for Utah geothermal power before 2032, which would conflict with the 1 January 2027 target for Southern California Edison's first 70 MW (Jefferies figure, relayed by Morpheus). This is an allegation (grade D). The May 2026 prospectus claims 290 MW of transmission, and a pathway retirement in June 2027 is also reported. No Fervo response was found, so delivery should be read as unresolved.
6.2 Seismicity and water
FINDING: Fervo states a four-step Induced Seismicity Mitigation Protocol without numeric thresholds, while Morpheus Research and Hagens Berman cite event counts that use different definitions (grades C and D).
SO WHAT: The two counts cannot be compared, and no independent seismic catalogue analysis has been done, so the seismicity risk cannot be sized.
NOW WHAT: Ask for the public microseismic catalogue and the protocol's numeric trigger levels, and have an independent seismologist review both.
Tim Latimer, Fervo's chief executive, told Fortune on 14 May 2026:
"Through four years of operations, we've had no challenges with seismicity associated with our projects."
It is a company claim. Morpheus alleges about 58 events of magnitude 2 or above through April 2026 and a magnitude 3.1 "Red Level" event about 24 days before its report, citing USGS data, and says Latimer once called seismicity a perception issue. Hagens Berman's release refers to about 58 "yellow-level" events. A magnitude threshold and a protocol alert level are different measures. Microseismic catalogues sit on the DOE Geothermal Data Repository, and we did not open them. On water, Fervo says it uses non-potable saline aquifers, a binary cycle and air-cooled systems. Morpheus alleges that Utah classed the use as non-consumptive, that PacifiCorp filed a protest and that the Utah Division of Water Rights said it would monitor (allegation; not independently verified). Morpheus alleges Fervo targets under 1% water loss at commercial scale; no data were provided.
6.3 Offtake concentration and contract terms
FINDING: Binding volumes (658 MW at IPO) and non-binding frameworks (Google's 3 GW through 2033) are reported side by side, and the customer-concentration percentage was not found (grades F and N).
SO WHAT: A buyer cannot plan around a framework, and a lender cannot size debt against one.
NOW WHAT: Count only signed, binding megawatts, and request penalty and performance-threshold terms for each contract.
Morpheus alleges that the Google 115 MW contract carries delay penalties above US$500,000 a week and default if output falls below 80% of contracted levels (allegation; no Fervo response found). Shell Energy's terms and any share of backlog held by a single customer were not found in the sources reviewed.
6.4 Replication, cost and equity risk
FINDING: Phase II sits beside Phase I, and Phase I cost of about US$7,000/kW compares with US$1,450–2,100/kW for gas combined cycle in Lazard's LCOE+ v19.0 (July 2026; grade C for the Fervo figure).
SO WHAT: Success at the same site does not show the approach works elsewhere, and the cost gap to gas means the case depends on price premiums and an unproven decline to US$3,000/kW.
NOW WHAT: Wait for a second site outside Utah and Nevada to produce power before treating replication as demonstrated.
MIT Technology Review noted on 6 October 2026 that Phase II sits next to Phase I, and its reporting paraphrases Shuvajit Bhattacharya of the University of Texas at Austin as saying that success at Cape would show Fervo can scale economically but not that the technology transfers easily.
A BloombergNEF gas build cost of US$2,157/kW (via IEEE Spectrum) is slightly above Lazard's range; the sources use different years, and Lazard's range is better documented ([CONFLICT], minor).
On equity, a Hagens Berman release of 13 October 2026 states that Fervo shares closed at US$14.33 on 6 October 2026, about 46% below the US$27 IPO price. Hagens Berman is an interested party, and it announced an investigation. Morpheus alleges a lock-up expiry in early November 2026 that could add about 205m shares; that date is allegation-sourced. Fervo's 10-Q, dated before the short report, says management was not aware of commitments or contingencies with a material effect. We offer no view on the share price.
6.5 Policy friction outside the United States
FINDING: The EU's Geothermal Action Plan was unpublished in mid-August 2026, German groups objected to the GeoBG on water protection, Indonesia is revising its tariff regulation, and India has no numeric target (grade C, secondary sources).
SO WHAT: Policy support outside the United States is slower and less specific, which delays any non-US replication case.
NOW WHAT: Model a permitting-delay case for every non-US project, and name the statute or fast track that would remove the delay.
The environmental group DUH and the industry bodies BDEW and VKU raised water-protection and operating-plan objections to Germany's GeoBG. For how EU rule-making timelines shape another clean-energy investment case, see our RED III RFNBO Targets 2030 analysis.
Table 8. Risk register with evidence labels · as of 22 October 2026
Risk | Evidence | Label | What would resolve it |
Transmission and delivery | 290 MW rights vs 384 MW contracted; 424B4 risk factors | Press, Filing | Firm interconnection agreement with in-service date |
Transmission pathway timing | "No viable pathway" before 2032, as alleged | Allegation | CPUC filing review; Fervo response |
Induced seismicity | Company protocol vs third-party event counts | Company, Allegation | Independent analysis of the microseismic catalogue |
Water | Non-consumptive classification; PacifiCorp protest (as alleged) | Allegation | Metered water-loss data at Cape |
Offtake concentration | 658 MW binding; 3 GW non-binding; penalty terms alleged | Filing, Allegation | Contract schedule by counterparty |
Replication | Phase II beside Phase I | Press | Second site outside Utah and Nevada producing power |
Cost competitiveness | US$7,000/kW vs gas US$1,450–2,100/kW | Filing, Press | Independent cost confirmation |
Equity and legal | Share price fall; investigation announced | Allegation | Fervo Q3 2026 results; court or regulator outcomes |
Well integrity | Blowout, 25 May 2026; no material effect reported | Company | Incident review in Q3 2026 filing |
Sector delivery | Eavor 0.5–2.0 MWe vs 8.2 MWe | Press | Eavor metered output statement |
Independent verification | No NREL, DOE, USGS or university validation of Cape found | Gap | Third-party reservoir and well-test review |
Source: Utility Dive (5 June 2026); MLQ.ai (5 May 2026); Fervo 424B4 and Form 10-Q; Morpheus Research (6 October 2026); Hagens Berman (13 October 2026); ThinkGeoEnergy (16 October 2026); IEA commentary (23 January 2026). The register ranks nothing; it lists risks and the evidence that would settle each.
7. Capital & Investment Implications
What does Fervo's US$421 million financing prove about bankability?
It proves that lenders were prepared to underwrite one project against specific contracts: Fervo had drawn US$212.3m of US$421.4m in Project Granite commitments by 30 June 2026. It does not prove that geothermal debt is repeatable on standard terms, because the loan terms and an independent cost basis were not available in the sources reviewed. Bankability here means the evidence a lender needs on resource, cost, delivery and revenue, and that evidence is still company-sourced.
7.1 What the financing proves and does not prove
FINDING: Project Granite is non-recourse debt with about US$421.4m of commitments, of which about US$212.3m, or 50.4% (GFJ arithmetic), was drawn at 30 June 2026 (Form 10-Q, grade F).
SO WHAT: Lenders, reported as Barclays and HSBC, accepted one project's contracts and construction plan; that is project-level confidence and not an asset-class standard.
NOW WHAT: Ask for the loan's conditions precedent, draw tests and reservoir-performance covenants before using it as a template.
The rest of the balance sheet shows a company in heavy build mode. Construction in process rose to US$1,235,160K at 30 June 2026 from US$789,571K at 31 December 2025. Cash stood at US$2,106.4m, up from US$461.8m at the end of 2025, after falling to US$280.8m in the quarter to 31 March 2026 and before the IPO proceeds arrived. Phase II debt is not disclosed in the sources read. The S-1/A estimates Phase II capital cost at US$940m and describes it as mostly unfunded, but it does not state, in the material we read, whether that figure is the total or the remaining spend. GFJ arithmetic (400 MW × US$5,500/kW) gives about US$2.2bn, which is not a disclosed number. We therefore show the two as a conflict and print neither as the total cost of Phase II.
Mitchell Smith, grid lead at the Mountain West Geothermal Consortium, said in IEEE Spectrum on 12 October 2026 that
“Fervo's first power achievement shows investors the size and scale of the opportunity.”
The statement is a view on investor interest; it does not address cost or reservoir life.
7.2 Cost benchmarks and their labels
FINDING: Every Fervo cost figure (US$7,000/kW, US$5,500/kW, US$3,000/kW) traces to Fervo, while Lazard puts conventional geothermal capital cost at US$5,135–6,635/kW and levelised cost at US$67–111/MWh ([SINGLE-SOURCE] for Fervo; Lazard LCOE+ v19.0, July 2026).
SO WHAT: Next-generation capital cost today sits above the conventional range, so returns depend on declines that have not yet been observed independently.
NOW WHAT: Run valuation at US$7,000/kW as the base case and treat US$5,500/kW and US$3,000/kW as stated targets.
Table 9. Cost benchmarks and evidence labels · as of 22 October 2026
Figure | Value | Label | Origin and note |
Fervo Phase I capex | About US$7,000/kW | [SINGLE-SOURCE] | Fervo (S-1/A); repeated by Canary Media, MIT Technology Review, Heatmap, IEEE Spectrum; no NREL, DOE or IEA confirmation |
Fervo Phase II capex target | About US$5,500/kW | [SINGLE-SOURCE] | Fervo Q2 2026 release; PacifiCorp reuses it |
Fervo long-term target | US$3,000/kW | [SINGLE-SOURCE] | Management "nth-of-a-kind" target |
Phase II total capex | US$940m (S-1/A, scope not stated) vs about US$2.2bn (GFJ arithmetic) | [CONFLICT] | Neither is a confirmed total |
Lazard geothermal, US, unsubsidised | US$67–111/MWh; US$5,135–6,635/kW; capacity factor 90%–80% | [2-SOURCE] with Rystad (capital cost), different bases | Inflation-adjusted from an earlier edition; no EGS split |
IRENA global weighted average, 2024 | US$60/MWh, down 16% from US$72/MWh; O&M US$125/kW-yr | [SINGLE-SOURCE] | Absolute values via a ThinkGeoEnergy summary; IRENA PDF not read |
Rystad conventional capex, global | About US$6,000/kW | [2-SOURCE] with Lazard | Overlaps Lazard's capital range |
NREL ATB 2024 EGS capex (2022 base year) | US$5,469/kW (near-field flash) to US$22,133/kW (deep binary) | [SINGLE-SOURCE] | Base year predates Fervo's 2025 drilling gains |
Current next-generation geothermal LCOE range (DOE Liftoff) | US$65–141/MWh; 2030 target US$60–70; 2035 target US$45 | [SINGLE-SOURCE] | Relies on Fervo data |
Forward cost targets | IEA about US$50/MWh by 2035; DOE US$45/MWh by 2035 | [2-SOURCE] | Targets and scenarios, not observed costs |
Wood Mackenzie, Cape Station | US$79/MWh, unsubsidised | [SINGLE-SOURCE] | 30 June 2025; not labelled LCOE by Wood Mackenzie |
Gas combined cycle | US$51–129/MWh; US$1,450–2,100/kW | [SINGLE-SOURCE] | Lazard only; upper bound reflects a 30% capacity factor |
Solar plus storage | US$61–156/MWh | [SINGLE-SOURCE] | Lazard only |
Offshore wind | US$105–167/MWh; US$5,600–7,100/kW | [SINGLE-SOURCE] | Lazard only |
US nuclear (Vogtle 3 and 4 basis) | US$175–255/MWh; US$12,300–17,570/kW | [SINGLE-SOURCE] | Lazard only |
Source: Fervo S-1/A and Q2 2026 release; Lazard LCOE+ v19.0 (July 2026); IRENA (June 2025) via ThinkGeoEnergy (5 August 2025); NREL ATB 2024; DOE Liftoff (October 2024); IEA (13 December 2024); Rystad Energy (20 May 2026); Wood Mackenzie (30 June 2025).
Methodology note. [2-SOURCE] means two independent documents give comparable figures; [SINGLE-SOURCE] means one origin, even if many outlets repeat it; [CONFLICT] means documented figures disagree. Where a figure is arithmetic by GFJ (for example 400 MW × US$5,500/kW), it is labelled "GFJ arithmetic" and not treated as a disclosed number.
For a comparable cost-range analysis of another capital-intensive clean-power technology,
see GFJ's Offshore Wind Investment Intelligence 2026–2030 and its Part II, and for the gas benchmark that geothermal is priced against in many markets, see LNG as a Transition Fuel 2026–2030.
Worked example: what a 50 MW GeoBlock must earn (illustrative)
This example applies the cost benchmarks above to a single 50 MW block, the size of a Phase II GeoBlock. It is an arithmetic illustration, not a forecast, and it is not a result Fervo has reported. Sourced inputs and illustrative inputs are separated.
Sourced inputs: capital cost of US$7,000/kW, US$5,500/kW or US$3,000/kW (Fervo, [SINGLE-SOURCE]); capacity factor of 90% or 80% (Lazard range); operating cost of US$125/kW-yr (IRENA, 2024 projects, via ThinkGeoEnergy). Illustrative inputs: block size of 50 MW (matches the planned Phase II block size) and a flat power price of US$100/MWh (a round figure at the bottom of the US$100–130/MWh firm clean power range; not a Fervo contract price).
Capital: 50,000 kW × US$7,000/kW = US$350.0m (GFJ arithmetic)Output at 90%: 50 MW × 8,760 h × 0.90 = 394,200 MWh a yearRevenue at US$100/MWh: US$39.4m a yearOperating cost: 50,000 kW × US$125 = US$6.25m a yearOperating margin: US$39.4m − US$6.25m = US$33.2m a yearSimple payback: US$350.0m ÷ US$33.2m = 10.6 years
Table 10. Simple payback for a 50 MW block, years (illustrative)
Capital cost | Block capital | Payback at 90% capacity factor | Payback at 80% capacity factor |
US$7,000/kW | US$350.0m | 10.6 | 12.2 |
US$5,500/kW | US$275.0m | 8.3 | 9.6 |
US$3,000/kW | US$150.0m | 4.5 | 5.2 |
Methodology note. Simple payback is block capital divided by annual operating margin, undiscounted, for a 50 MW block at a flat US$100/MWh and US$125/kW-yr operating cost. Capital costs and capacity factors are the ranges sourced in Table 9; the block size and price are illustrative. Debt service, tax credits, taxes, reservoir decline and re-drilling are excluded. The table is an arithmetic illustration, not a forecast.
What it shows. At today's reported cost, a block repays its capital in roughly 11–12 years on these inputs, which uses most of a 15-year contract term before any return. At the US$5,500/kW target the figure falls to 8–10 years. The result also excludes debt service, tax credits, taxes, reservoir decline and re-drilling, each of which moves the answer. It shows why the cost-decline path, and not the contracted price alone, drives the investment case.
7.3 The Geothermal Bankability Scorecard
FINDING: Of six bankability dimensions, only construction and commissioning rests on a primary company release (F); the resource dimension rests on company data alone (C) and transmission is contested (D).
SO WHAT: The grades describe the quality of the evidence, so a low grade means a gap in the record and not a poor project.
NOW WHAT: Re-grade each dimension after every quarterly filing and fund only when the three weakest dimensions reach at least F with a third-party check.
Table 11. Geothermal Bankability Scorecard: Cape Station · evidence grades as of 22 October 2026
Dimension | What is graded | Evidence available today | Grade and date |
Resource and drilling risk | Reservoir temperature, flow, recovery factor, drilling cost | Company data only; not independently verified; recovery factor disputed | C D · 13 April 2026 |
Construction and commissioning | Schedule and cost against plan | GeoBlock 1 delivered one day early (company); later blocks pending | F · 1 October 2026 |
Capital cost and financing | Capex per kW, debt, equity | Filings show financing; no independent cost confirmation | F C · 12 August 2026 |
Contracted revenue and delivery | PPA price, tenor, penalties, counterparty | 658 MW binding; prices mostly undisclosed | F N · May 2026 |
Operating performance | Net MW, uptime, decline | Project Red 614 days; Cape about three weeks | C · 13 April 2026 |
Grid, permitting, environment | Interconnection, permits, seismicity, water | Transmission unresolved; permits supportive | D F · 6 October 2026 |
Source: GFJ assessment of Fervo filings and releases, Utility Dive, BLM testimony (16 December 2025) and third-party material labelled in Section 6.
Methodology note. The scorecard is structured analyst judgement, not a quantitative model. Each of six dimensions is graded V / F / C / D / N by the quality of its evidence:
V needs an independent source (for example a regulator, laboratory or audited third party),
F a primary filing or company release,
C an uncorroborated company claim,
D a contested or allegation-only claim,
N means no source was found. Criteria: source independence, primary versus secondary origin, date, and whether the claim describes an observed result or a forecast.
Dimensions are not weighted and no overall score is given, because the research for this report holds no comparable data across developers.
7.4 Contract and buyer economics
FINDING: Google's 396 MW agreement carries an option on about 600 MW more by June 2030, while its 3 GW framework through 2033 is non-binding; PacifiCorp's 2027 planning model uses a 500 MW proxy resource at about US$5,500/kW, falling to about US$3,000/kW by 2036, based on Fervo's own estimates (grade C).
SO WHAT: Buyers are pricing in cost declines that Fervo, and not an independent party, has estimated.
NOW WHAT: A buyer should seek price protection that adjusts if delivered cost or output differs from the planning assumption.
Lucia Tian, Google's director of advanced energy technologies, said in a Wall Street Journal interview reported by Canary Media on 1 September 2026 that Google knows the project
“will be a foundational building block of power generation for a data center presence in Utah.”
On price, Rystad Energy puts firm clean power at US$100–130/MWh, second-hand through Fervo's S-1/A, and the IEA notes recent contracts up to about US$130/MWh against US$30–60/MWh earlier ([2-SOURCE] on the upper bound).
For how price premiums for scarce clean products are set in another market, see The SAF Price Premium, Part I.
8. Future Scenarios & Forecast, 2026–2035
What milestones will show whether next-generation geothermal scales by 2030?
Three dated milestones matter most. First, Cape Station's remaining Phase I blocks are due online by 1 January 2027. Second, 12 months of stable Cape output from 30 September 2026 would give the first usable operating record. Third, Phase II (400 MW) and Google's 396 MW agreement are due in 2028. A second site outside Utah and Nevada producing power would test replication. Company targets, including 1.1 GW by 2030, are targets, not forecasts.
The scenarios below describe what would have to be observed for each outcome. They carry no probabilities and no price forecast. Company targets appear as targets: about 100 MW of Phase I by 1 January 2027, 400 MW of Phase II and Google's 396 MW in 2028, and 1.1 GW by 2030 across Fervo's portfolio, not at Cape alone (Fervo Q2 2026 release, 12 August 2026). MIT Technology Review (6 October 2026) reports a target of 1 GW by the end of 2030 and 5 GW about five years later, and the two 2030 figures have not been reconciled. The IEA's 800 GW deployable by 2050 and its cost of about US$50/MWh by 2035 are IEA scenario figures, not forecasts GFJ endorses.
For another long-horizon investment case with the same discipline between targets and evidence, see Green Molecules Economy 2035.
Table 12. Scenario triggers, 2026–2030 · no probability weights
Scenario | What it describes | Observable triggers |
A. Replication demonstrated | Cape performs, costs fall toward the stated targets, and a second site produces power | Phase I complete by 1 January 2027; 12 months of stable Cape output; Phase II on schedule; a site outside Utah and Nevada produces power |
B. Cape-only success | Cape performs, but costs and timelines do not improve at new sites | Cape output holds; cost declines stall near Phase I levels; interconnection delays push Phase II; new sites take longer |
C. Reservoir or delivery shortfall | Cape underdelivers on temperature, output or grid access | Thermal decline exceeds Fervo's model; output falls below contract thresholds; transmission pathway not delivered |
Source: GFJ analysis of Fervo filings and releases, the Google agreement (1 September 2026), PacifiCorp planning materials (Deseret News, 14 October 2026) and the risks documented in Section 6.
Methodology note. Scenarios A–C are conditional narratives with observable triggers. They carry no probability weights because no evidence in the sources reviewed supports any particular weighting. Triggers come from disclosed targets and dates (Fervo filings, Google agreement, PacifiCorp IRP) and from risks documented in Section 6.
Table 13. Dated watch-list · as of 22 October 2026
Date | Event | Why it matters | Label |
Early November 2026 | Reported lock-up expiry on Fervo shares | Date comes from an allegation; Fervo has not stated it | D |
Date not found | Fervo Q3 2026 earnings | First company update after the short report and GeoBlock 1 | N |
End-2026 | Quaise Oregon flow test (plan) | First field data for superhot-rock drilling | C |
1 January 2027 | Phase I blocks 2 and 3 due online | Tests repeat construction and the transmission path | F |
2027 | PacifiCorp integrated resource plan models a 500 MW proxy resource | Shows whether a utility plans on Fervo's own cost estimates | C |
2028 | Phase II (400 MW) and Google 396 MW due online | The replication test at scale | F |
2030 | PacifiCorp proxy resource first eligible; Fervo 1.1 GW portfolio target | Marks the end of the 2026–2030 window | C |
Source: Hagens Berman and Morpheus Research (lock-up date, allegation-sourced); EnergyConnects (July 2026); Fervo release (1 October 2026); Deseret News (14 October 2026); Fervo Q2 2026 release (12 August 2026).
Methodology note. Dates come from company filings and releases or from the third-party sources named under the table, and each label is an evidence grade from Section 1.2. Events with grade N have no confirmed date. The list is chronological, is not ranked and carries no probabilities.
Continue reading part II - https://www.greenfueljournal.com/post/next-generation-geothermal-investment-2026-2030-what-fervo-s-cape-station-proves-part-ii