Space Power: the financing landscape from solar arrays to orbital utilities

Star Catcher is the closest thing in orbit to a merchant utility: satellites that collect and re-beam concentrated solar spectrum onto customers' existing panels, sold as power-purchase agreements. The company reports 7 signed PPAs — including Starcloud, Loft Orbital (BusinessWire, Nov 2025) and Astro Digital (Sep 2025) — described as "tens of millions" of ARR through end of decade, against a claimed >$3B qualified pipeline. Every one of those figures is company-stated and unaudited; treat them as marketing until an on-orbit service window opens.

What is independently verifiable is the hardware ladder: a 100m ground beaming demo at EverBank Stadium (Mar 2025), then a 1.1 kW optical power-beaming record at Space Florida's Launch and Landing Facility in Nov 2025 (PRNewswire) — beating the 800 W over 8.6 km that DARPA's POWER program set in mid-2025 (DARPA) — followed by the oversubscribed $65M Series A in May 2026. Beaming onto panels customers already fly removes receiver-side capex and adoption friction, which is the segment's best structural unit-economics hack.

The credit frame: history rhymes. Solaren signed a 200 MW space-solar PPA with PG&E in 2009 and never flew hardware. A pre-revenue PPA contingent on an unproven orbital network is an option with zero recovery value, not contracted backlog — the right analogy is a pre-construction terrestrial PPA on a developer with no completed plant. The book is the most credit-analyzable asset in the cohort precisely because it can be marked to that standard. The de-risking gate is the first orbital service window, which the company targets in the 2026–27 span.

Aetherflux is the sector's purest single-demo bet. Disclosed financing is a $10M founder seed (Baiju Bhatt) plus a $50M Series A (Apr 2025, Index and Interlagos with Breakthrough Energy, a16z, NEA — TechCrunch); March 2026 reporting puts total capital collected at ~$80M, implying interim money beyond the disclosed rounds. The reported $250–350M Series B at a ~$2B valuation, led by Index, was in talks and not confirmed closed as of this writing (DCD) — a valuation mark that rests entirely on Mission 1, a 2026 laser-beaming demo flying on a purchased Apex bus via Falcon 9 rideshare.

Three underwriting observations. First, the capital-light architecture (buy the bus, ride share the launch) is deliberate and smart — but it means the company's only hard asset is one spacecraft and the team. Second, the DoD OECIF award validates interest, not demand: it is a development grant of undisclosed size, not procurement. Third, the company has already pivoted its lead narrative toward orbital data centers ("Galactic Brain," targeted operational Q1 2027 per DCD) — which both broadens the story and signals that beam-to-Earth alone was not carrying a $2B mark. If Mission 1's end-to-end laser link fails, there is no PPA book, no backlog, and no second spacecraft to point to. That is the definition of binary risk, and it is why on-orbit performance cover for novel beaming payloads is effectively unbuyable today.

Space nuclear is the one layer where government demand is real and growing: NASA's Fission Surface Power program was reset in July 2025 to a ≥100 kW, ≤15 t lunar reactor with a launch-ready target of end-2029, with the flight-unit partnership competition expected in early 2026 (SpaceNews). The cohort sorts by contract validation and fuel access, not by round size:

• Zeno Power — the most contract-validated startup in the segment: >$60M of DoD/NASA awards (company-stated; a $15M 2022 DoD award is corroborated in USAspending) before its $50M Series B (May 2025, Axios), >$70M raised in total. Its real moat is dual-sourced fuel: DOE Sr-90 plus an Orano agreement unlocking Am-241 (Orano, Sep 2025). First full-scale RPS demo targeted 2026, commercial deliveries 2027.
• X-energy — the strongest balance sheet touching the lunar reactor competition: a $700M Series C-1 (Feb 2025, anchored by Amazon's Climate Pledge Fund) followed by a $700M Series D (Nov 2025), roughly $2.5B cumulative. Space is optionality on a terrestrial SMR story; its IX joint venture with Intuitive Machines holds one of the three FSP Phase 1 design contracts, and its TRISO-X Oak Ridge line is the only commercial TRISO fuel plant at scale.
• Standard Nuclear — the corrected datapoint that re-tiers the fuel layer: after buying Ultra Safe Nuclear's TRISO assets out of Chapter 11 for $28M (Oct 2024), it closed a $140M Series A in January 2026 (TechCrunch) and was selected for DOE's fuel pilot program. Its claimed $100M of 2027 fuel sales are non-binding LOIs — not backlog — and ownership of USNC's EmberCore space radioisotope IP post-auction still needs confirmation before underwriting anything against it.
• Antares Industries — $96M Series B (Dec 2025) split $71M equity + $25M debt (SpaceNews); venture debt in a pre-demonstration nuclear company is an aggressive precedent. DOE HALEU feedstock allocation and Reactor Pilot Program selection (Aug 2025) make fuel access part of the asset.
• Avalanche Energy — micro-fusion, $40M Series A (2023) plus $29M led by RA Capital (Feb 2026, Bloomberg), partly matching a $10M Washington State grant. Physics risk dominates, but its FusionWERX test facility is a revenue line and rare downside protection for a fusion venture.
• Incumbent exposure — BWXT ($2.7B FY2024 revenue, NYSE: BWXT) is the de facto fuel-and-reactor subcontractor to whichever team wins the NASA flight unit; the DRACO nuclear-thermal program it served was effectively terminated by the FY26 budget request, a reminder that appropriation risk applies even to marquee programs. JETSON design contracts (Lockheed $33.7M, Westinghouse ~$17M, Intuitive Machines $9.5M — SpaceNews) round out the AFRL channel.

The structural credit point: launch authorization for any commercial space reactor runs through the NSPM-20 presidential process, fuel is government-allocated (Pu-238 scarce; HALEU/TRISO via DOE), and Price-Anderson-style indemnification does not cleanly cover commercial space reactors. Fuel-supply agreements are the bankable collateral in this layer — more so than design contracts.

Space photovoltaics. III-V multijunction space cells run on the order of hundreds of $/W (a commonly cited ~1,000x terrestrial; review: Frontiers in Physics) versus terrestrial silicon at $0.20–0.40/W. Solestial targets ~$10/W silicon space modules at MW scale — a 10–30x cost wedge at 60–70% of III-V efficiency (Si ~20–26% vs 3J/4J 30–35%), which wins in LEO where radiation life is short anyway. The catch is the revenue ceiling:

The conclusion drops out of the geometry: GEO microwave needs kilometer-scale structures (hence the robotics/in-space-assembly dependency), LEO laser needs none — which is why essentially all 2024–26 venture money chose lasers. And the two best unit-economics hacks in the segment are both receiver-side: Overview beams near-IR onto existing terrestrial solar farms; Star Catcher concentrates spectrum onto existing satellite panels. Neither requires the customer to build anything. TAM figures, for the record, diverge 10x across the named research firms (MarketsandMarkets has SBSP at $4.7B in 2030 → $6.8B by 2040; Mordor and Grand View are far lower) — the spread itself is the signal that no commercial orbital kWh has ever been sold. We quote them for completeness and use none of them for underwriting.

Demo risk is uninsurable; bus risk is not. Launch insurance and standard first-year bus perils are purchasable for every spacecraft in this sector. What has no actuarial base is on-orbit performance cover for novel beaming payloads — expect the technology risk on Aetherflux Mission 1, Star Catcher's first nodes, Overview's 2028 demo and JAXA's OHISAMA to ride uninsured through roughly 2028, with underwriters quoting only conventional hull perils around it.

Mark PPAs as options, not revenue. Star Catcher's 7 PPAs and the Meta–Overview agreement (up to 1 GW of early-access capacity by end of decade, terms undisclosed and almost certainly contingent — SpaceNews, Apr 2026) mirror terrestrial pre-construction PPAs, and the 2009 Solaren–PG&E precedent says exactly how much recovery to assume on failure: zero. What the hyperscaler entry does change is counterparty quality — an offtaker like Meta materially improves the financing story versus 2023, when DoD grants were the only demand signal.

Beam liability and social license are total-loss scenarios for specific names. Laser and microwave beams crossing airspace raise third-party liability questions spread across FAA, FCC and ITU jurisdiction; Reflect Orbital additionally faces organized opposition (DarkSky International) and an unresolved FCC application for 4,000 mirrors — regulatory denial would not impair its model, it would end it. Nuclear adds the NSPM-20 launch-authorization process and bespoke launch-nuclear cover, since Price-Anderson-style indemnification does not cleanly extend to commercial space reactors. Fuel supply is the credit-critical input across the cohort — Zeno's dual-sourcing is the model.