Capital Stacks For Atoms: how atom-based companies are financed, with the Fervo case

By Software Synthesis (Akash Bajwa) · Monday, October 12, 2026.

The essay starts from a shift in capital allocation over recent years: money is moving from businesses built on bits to businesses built on atoms. According to the author, the trillion-dollar AI build-out has exposed the physical bottlenecks of inference. Investors are directing capital toward those bottlenecks in public markets, private markets and early-stage venture capital.

As a theoretical foundation, the author cites a piece Brett Bivens wrote last year, "The Rise of Production Capital". It describes the capital structure needs of atoms companies, which span energy, manufacturing, space and robotics. Examples of companies where debt plays an ever larger role include Zenobē (transport fleets), CoreWeave (digital infrastructure), Base (energy storage), Hadrian (manufacturing), Twelve (clean fuels) and ISAR (aerospace).

The cited piece argues that, for companies deploying valuable physical assets, asset-backed lending works as a wedge that opens up project-level and equipment-level financing. This lets them grow at venture speed while consuming far less equity capital. These businesses are built and valued on bundles of cash flows, not on revenue multiples.

It describes the optimal progression as follows. First, venture equity for R&D and product development. Next, targeted asset-backed debt once the technology proves commercial viability. Later, warehouse facilities or securitization structures to finance several deployments at once and recycle capital without going back to venture. Securitization turns illiquid hardware into tradable securities with predictable cash flows. The author says he has been thinking about this for some time, but only in theory, and that recent examples show how capital stacks evolve as a technology climbs the technology readiness levels (TRL).

The main case study is Fervo, in the energy sector. The company produces geothermal power where conventional geothermal does not work. It drills horizontal wells into hot, dry rock and fractures it to create a reservoir, using techniques from shale oil and gas. Its pitch is repeatability: standardized GeoBlocks, drilled in the same basin, following the same learning curve that made shale cheaper. CEO Tim Latimer frames it in the IPO prospectus letter as treating each geothermal plant as a repeatable product rather than a complex one-off project. Fervo was founded in 2017 by Tim Latimer and Jack Norbeck.

The flagship site is Cape Station, in Beaver County, Utah. The pilot, Project Red in Nevada, produced 3.5 MW in July 2023, the first enhanced geothermal system with horizontal wells. Phase I is about 100 MW across three GeoBlocks, contracted with Southern California Edison, Shell Energy and community choice aggregators. GeoBlock 1 declared commercial operation on September 30, 2026, one day ahead of the contractual date, with 33 MW net. Phase II targets about 400 MW in 2028, with roughly $2.2 billion of capex through 2028.

In September, Google signed a 396 MW PPA, with an option on roughly 600 MW more before June 2030. In the second quarter, Fervo had 658 MW of binding PPAs, with about 15 years remaining and a $7.2 billion revenue backlog. With Google, binding PPAs passed 1 GW in October.

On costs, Phase I comes in around $7,000/kW, the Phase II target is about $5,500/kW and the long-term target is $3,000/kW. For reference, an average gas plant built in the US in 2025 cost $2,157/kW. The ninth well using the current design reached about 19,500 feet in 21 days, a company record. The author stresses that Fervo is a construction company: its second-quarter revenue was $113,000 and its capex $226.5 million.

The author breaks Fervo's financing into stages.

Stage 1, technology risk (2017-2023). The stack was venture equity plus non-dilutive public money. Series A, B and C totaled about $177 million. DOE and ARPA-E awarded about $7.9 million across four grants between 2018 and 2022.

Stage 2, the bridge years (2023-2025). Cape Station needed cash before it could take on bank debt, and Fervo used every instrument available. Between the pilot and access to broader credit markets lies a band of expensive, bespoke capital:

- 2023 convertible note: $10 million from a holder of more than 10%, with a 3.1% coupon and a 15% conversion discount. Ten months later it converted into $12.0 million of Series D-2. - Series D-1 (February 2024): $3.52 per share, the same as Series C despite Project Red's results. The author presents this as the cost of equity before Cape had a lender. - XRC bridge (August 2024): $100 million, later expanded to $145.6 million, from a "micro-rural" construction lender, at prime + 3.5% with an 11% floor. It included a prepayment premium and sponsor guarantees, among them a completion guarantee. - Mercuria term loan (November 2024): $40 million, later expanded to $100 million, plus an $80 million letter of credit facility. Pricing: SOFR + 0.10% + 7.50% with a 3% floor, a 1.5% upfront fee and a 2% commitment fee on undrawn amounts. Covenants include a debt-to-equity-contributions ratio of no more than 0.6x.

Stage 3, project preferred at two prices. There were two tranches of preferred in Cape Phase I, a few months apart and at very different prices. Breakthrough Energy Catalyst put in $100 million in June 2025 with a 5% annual compounding preferred return and cumulative distributions of about $139 million on an amortizing profile through 2041. Centaurus put in $75 million of junior preferred in October 2025, with a cash waterfall up to a 12.5% IRR flip (12% if no term debt is raised) and then a $5/MWh royalty starting 15 years after final completion. It also included warrants on 2.55 million shares at $7.34. According to the piece, the Catalyst investment is project-level preferred with a priority dividend and return of capital, and cash goes first to preferred distributions. The author concludes that the gap between 5% and 12.5% plus warrants is the price of seniority.

Stage 4, bankability (March 2026). Project Granite is the moment the asset became bank-financed infrastructure. It is a $421.4 million package: a $309 million construction-to-term loan, a $61 million tax credit bridge and $51 million in letters of credit. The coordinating lead arrangers were Barclays, BBVA, HSBC, MUFG, RBC and Société Générale, alongside BofA, J.P. Morgan and Sumitomo Mitsui Trust.

The terms were as follows:

- Construction and term loans: SOFR + 3.00%, stepping up to 3.125% and then 3.25% from the third and fourth anniversaries. - Tax credit bridge: SOFR + 2.375%. - Commitment fees: 30% of the applicable margin. - Tenor: five-year mini-perm, maturing in March 2031. - Sizing: the lesser of a minimum 1.50x DSCR on the P90 production forecast and 75% of total project cost.

Compared with Mercuria's 7.60% over SOFR, the author calculates spread compression of about 460 basis points in 16 months.

This summary does not cover the rest of the essay. For the full text, the email points to the web version: https://www.akashbajwa.co/p/capital-stacks-for-atoms.