Last Energy

Phase A — Understand the business

Lens 1 · Company Overview

Last Energy is a private, Washington-D.C.- and Austin-founded developer of micro-modular pressurized-water nuclear power plants, established 2019 by Bret Kugelmass as the for-profit spin-out of his Energy Impact Center; it launched publicly in February 2020 with a $3M seed led by First Round Capital. The product is the PWR-20 — a 20 MWe / 80 MWth pressurized-water reactor sold not as a machine but as electrons. This is the load-bearing fact of the whole business: like an independent power producer (IPP), Last Energy builds, owns and operates the plant on the customer's site and sells the output under long-term Power Purchase Agreements (PPAs). It explicitly does not seek government equity to develop plants — a deliberate contrast to NuScale (DOE-cost-shared) and TerraPower/X-energy (DOE ARDP grants).

Who the customers are: industrial baseload buyers who need firm 24/7 power behind-the-meter or via private wire — data centres, manufacturers, steel mills, special economic zones. Roughly half the disclosed commercial agreements are earmarked for data-centre developers (39 of an ~80-unit claimed book). The AI-data-centre power crunch is the explicit demand thesis — this is an AI-infrastructure / power name wearing an energy-beat label.

Contract structure: long-dated PPAs (the Poland KSSE deal is quoted as >$4.3B of electricity sales over the contract lifetime, i.e. ~20-30yr revenue, not booked backlog). Take-or-pay / offtake terms are not publicly disclosed — a material unknown.

One-line business model: a fabless-style nuclear IPP — standardize one small reactor, mass-produce it, own the fleet, sell power. "We don't think in ones and twos, we think in tens of thousands," Kugelmass says; the stated 15-year target is 10,000 units.

Lens 2 · Supply Chain

The differentiating design choice is radical off-the-shelf-ness, so the supply chain is deliberately boring — that is the moat claim (Lens 3). Mapping upstream → Last Energy → customer:

• Fuel (upstream): standard full-length 17×17 PWR fuel assemblies enriched to <4.95% U-235 — the same LEU that fuels the ~300 operating LWRs supplying ~10% of global electricity. Suppliers are the existing Western fuel-fabrication base (Westinghouse, Framatome, Global Nuclear Fuel) — not named by the company, but crucially not HALEU. This sidesteps the 19.75% HALEU bottleneck (single-sourced historically from Russia/Rosatom; US supply via Centrus [NYSE: LEU] only now ramping) that gates Oklo, X-energy and TerraPower. Last Energy says it has secured its first full core load, arriving September 2026, and each core carries ~6 years of fuel.
• Reactor / pressure vessel: design lineage is a 1960s PWR originally developed for the NS Savannah nuclear merchant ship, adapted into a permanently steel-encased, sealed core (~1,000 tons of steel per core, ~$1M of steel) that is never opened on-site — the encasement doubles as decommissioning waste containment. Vessel/steel fabrication partner not disclosed.
• Balance of plant: "few dozen" factory-made rectangular modules — modular steam plant, air-cooled condensers (closed-cycle air cooling, no water source required), "plug-in-ready" electrical skids. Manufacturing philosophy borrowed from oil & gas modular fabrication.
• Assembly: modules trucked to site, assembled on <0.5 acre (114 sq ft "nuclear island"); claimed 24-month deliver-and-assemble cycle.
• Strategic supplier tell: The Haskell Company (a large US engineering/construction/fabrication firm) is both a Series C investor and namesake-adjacent to the Texas site — a strategic EPC relationship hiding inside the cap table.

Chokepoints: (1) the factory — there is no disclosed serial-production plant yet; "design once, build many" is a plan, not a running line; (2) project-finance debt — equity funds development, but each plant is a capital asset the fleet must debt-finance (EXIM's $103.7M debt LOI for one Welsh unit is the tell of how capital-heavy each unit is); (3) first-of-a-kind engineering — no unit has ever operated. Names or it didn't happen: the supplier names above are largely inferred/undisclosed — a genuine gap in a company this pre-commercial.

Lens 3 · Competitive Advantages (moats)

Last Energy's moat pitch is "we removed the two things that kill nuclear timelines: novel fuel and novel physics." Assessed durably:

• Fuel-supply moat (real, and the strongest card). Using licensed <5% LEU means no HALEU queue, no new enrichment license, no fuel-qualification campaign. Against Oklo (fast-spectrum HALEU), X-energy (TRISO HALEU), Kairos (fluoride-salt), this is a years-of-schedule advantage that is hard to argue with. Durable because it's structural to the design.
• Proven-physics moat (real but shared). A PWR is the most-operated reactor type on earth (~300 units) — regulators and insurers understand it. But this cuts both ways: NuScale, Rolls-Royce SMR, Holtec and Westinghouse AP300 are also LWR/PWR-based, so "PWR-ness" is not unique — the micro-scale + factory + own-and-operate packaging is.
• Business-model moat (differentiated, unproven). Selling electrons not reactors collapses the customer's decision from "build a nuclear plant" to "sign a power contract," and keeps Last Energy off the customer's balance sheet. If it works, it's a genuinely different go-to-market from every reactor vendor. But it means Last Energy carries the capital and operating risk of an unproven fleet — the moat is also the millstone.
• Bargaining power: weak today. A pre-revenue developer with zero operating units has little leverage over fuel fabricators, EPC firms, or lenders; its leverage over customers rests entirely on the AI-power scarcity making 20 MW firm blocks precious. PPAs signed with special economic zones and industrials suggest customers see optionality value, but these are cancellable if no plant is delivered.
• What is NOT a moat: IP. A 1960s ship-reactor derivative is not a patent fortress; the edge is execution and standardization speed, not proprietary science. Ground-truth: patents/ on the shelf is empty.

Verdict on moat: the fuel/physics advantage is real and buys schedule; the enduring moat only exists if they become the first to actually operate a fleet and ride a manufacturing learning curve nobody else has. Pre-first-power, the moat is a thesis.

Lens 4 · Segments

No reported financial segmentation exists — pre-revenue, private, segments.csv empty. The only meaningful decomposition is the order-book pipeline by geography/end-market (all ``, all MOU/PPA-stage, none delivered):

Trend & credibility: the book has grown (34 units/$18.9B in 2023 → ~55 units early-2024 → ~80 claimed) but the quality is unchanged — these are agreements-to-buy-power-if-built, not financed firm orders, and zero have converted to a delivered MWh. The single most important segment number is therefore units operating: 0.

---

Phase B — Measure performance

Lens 5 · Funding & valuation trajectory (+private swap for "Earnings Result")

No earnings. The scoreboard is the capital stack:

• Disclosed equity to date ≈ $163M.
• ⚠ Provenance conflict — surfaced, not resolved: aggregators state "raised $324M … 21 investors" (adding names like Chaucer Global, Dysruptek, Endowment Arm). This ~$161M gap over the summed announced rounds is unexplained — likely counts the EXIM $103.7M debt LOI, undisclosed bridge/grant capital, or double-counts. I do not treat $324M as clean equity; the defensible figure is ~$163M announced equity + up to ~$104M contingent debt LOI.
• Valuation: n/a — not disclosed. No round has published a pre/post-money mark; secondaries are listed on EquityZen/Forge/Nasdaq Private Market but no reliable clearing price surfaced. Any valuation number would be fabrication — I decline to invent one.
• Burn signal: having raised ~$163M over 6 years and needing external debt (EXIM) for a single first unit tells you the equity is development capital, not fleet capex. The Series C is explicitly to fund the PWR-5 pilot + first commercial units, not the 10,000-unit dream.

Lens 6 · Founder/leadership signal (+private swap for "Earnings Calls")

There are no earnings calls; the substitute is the founder's unusually large public corpus. Kugelmass hosts "Titans of Nuclear" and conducted 1,500+ interviews with nuclear operators/experts before founding the company. Sentiment/positioning over time:

• Consistent thesis: nuclear's problem is cost and schedule, not physics or safety; the fix is standardization + manufacturing + deregulation. This has been stable since 2017 — a point in his favor (no strategy drift).
• Escalating combativeness toward regulators: rhetoric moved from "build abroad because US licensing is too slow" (2023-24) to suing the NRC (Dec 2024) — a hardening tone (Lens 10/13).
• Recurring phrases: "off-the-shelf," "we sell electricity not reactors," "tens of thousands," "proportionality." What's conspicuously absent from his talk track: a firm price per MWh — he explicitly won't commit to one, arguing nuclear can't ride the same learning curve as solar because of fixed regulatory cost. For a company whose entire pitch is cheap firm power, refusing to quote a cost is the single most important thing he stopped saying / never said.

Lens 7 · Cap table quality & comps (+private swap)

Syndicate quality (the IPO-proximity tell): the Series C is led by Astera Institute (Eric Schmidt-linked science funder) with strategics (Haskell = EPC; Woori = Korean financial/industrial; Galaxy = crypto/AI crossover) and founder-adjacent funds (Gigafund, JAM Fund/Calacanis). Crucially absent: any tier-1 IPO-signal crossover fund — no Fidelity, T. Rowe, Coatue, Wellington, BlackRock. Per the +private playbook, the absence of a crossover mark is a low-IPO-proximity signal. This is a conviction/mission cap table, not a pre-IPO cap table.

Comps (peer microreactor/SMR developers — public multiples are near-meaningless as all are pre-revenue, so I show stage/tech/fuel/anchor instead; P/E, EV/Sales, EV/EBIT = n/a / not meaningful pre-revenue):

Read: every public microreactor comp is a pre-revenue, sentiment-driven name that sold off 27-30% in H1 2026 — the sector re-rated down even as the AI-power narrative held. Last Energy's private mark is untested against that repricing. Its LEU-vs-HALEU column is the one place it screens better than the field.

Lens 8 · Catalyst / event history (+private swap for "Price Catalysts")

No stock, so the tape is the milestone cadence that has (de-)risked the story:

• Feb 2020 — public launch, OPEN100 open-source design + $3M seed. (narrative genesis)
• 2022 — Poland (Enea, Legnica SEZ) + Romania (RATEN) JDAs. (order-book genesis)
• Oct 2023 / Mar 2024 — physical prototype displayed in Poland and at CERAWeek Houston. (tangibility proof — a mock-up, not an operating unit)
• Aug 2024 — $40M Series B; ~80-unit book claimed.
• Dec 2024 — EXIM $103.7M debt LOI for first Welsh unit; NRC lawsuit filed (Dec 30). (financing + regulatory-arbitrage fork)
• Feb 2025 — UK ONR admits Wales project into nuclear site licensing — first new commercial nuclear site into UK licensing since Torness (1978). Strongest real milestone to date.
• May 2025 — Texas Haskell County 30-unit / 600 MW data-centre plan; ERCOT interconnection filed; NRC early-site-permit pre-application.
• Aug 2025 — UK Preliminary Design Review (PDR) completed with ONR + Environment Agency.
• Dec 2025 — $100M+ Series C (Astera).
• 2026 (in progress) — RELLIS PWR-5 pilot (Texas A&M, DOE Reactor Pilot Program): reactor pits complete & lined, building rising, components fabricated, fuel ordered; criticality demonstration targeted 2026; first full fuel core arriving Sep 2026. ← the decisive upcoming catalyst.

Pattern: the market (of private capital + counterparties) reacts to regulatory admissions and financing LOIs, not to prototypes. The next re-rate hinges on a working reactor, not another PPA.

(+private add) · Traction & unit economics

• Revenue run-rate: $0. No unit online; financials.csv empty.
• Pipeline: ~$18.9B lifetime PPA value across 34 firmed PPAs (UK+Poland), ~80 units claimed, ~600 MW Texas. Contracted MW delivered: 0.
• Unit economics: not disclosed. No LCOE, no $/kW capex, no PPA strike price is public — the founder actively declines to quote price. Only a component tell exists: ~$1M of steel per core, which is a fraction of true all-in per-plant cost (vessel internals, fuel, BoP, EPC, grid, licensing). Any full capex/MWh figure here would be `` with too little to anchor — n/a.

---

Phase C — Judge people & books

Lens 9 · Management

• Bret Kugelmass (Founder/CEO). BS math (Stony Brook), MS mechanical/robotics (Stanford); ex-Nanosolar mechatronics, Panasonic EV lab, NASA Goddard (Mars-rover control unit); founded Airphrame (drone fleets), sold 2017; then founded Energy Impact Center (2017) and Last Energy (2019). Archetype: technical founder-evangelist / category entrepreneur, not a nuclear-industry lifer or an operator who has built a power plant.
• Track record — what he's actually built: a sold drone startup and a genuinely influential nuclear-media platform (Titans of Nuclear, 1,500+ interviews). What he has not built: an operating nuclear reactor. The gap between narrative fluency and operating experience is the central management question.
• Skin in the game: founder-led, presumably large ownership (undisclosed); Gigafund/JAM Fund backing him across multiple rounds signals aligned founder-friendly capital.
• Capital allocation: disciplined narrative (no government equity, standardize-and-manufacture) but unproven execution — the test is whether Series C dollars convert to a critical reactor in 2026.
• Red flags (soft): promotional intensity (see Lens 10); a regulatory-arbitrage posture (build abroad, sue the domestic regulator) that a charitable read calls "pragmatic" and a skeptical read calls "can't clear US licensing so route around it."
• Bench depth: not surfaced. No CTO/COO/chief nuclear officer profile appeared in public sources — a gap for a company claiming a 2026 criticality. For a first-of-a-kind reactor, who has operated one before matters more than the founder's story.

Lens 10 · Forensic / promoter-risk audit + Regulatory findings

There is no audited financial statement to forensically dissect (private, no filings). The honest forensic frame for a pre-revenue promoter is claim-integrity and execution risk:

• Headline-inflation risk (elevated). "$18.9B pipeline / largest in the world" is lifetime PPA revenue on undelivered MOUs, presented adjacent to "0 reactors built." "80 units," "10,000 in 15 years," "24-month delivery" are aspirations stated as near-facts. Wikipedia itself carries a "contains promotional content" maintenance flag (Jan 2026) — an unusual, citable external tell.
• Cost-omission (the big one). A firm-power company that won't publish a price is asking to be underwritten on faith. Absent LCOE, the economic claim is unfalsifiable.
• Design-heritage risk. A 1960s NS-Savannah-derived PWR is "proven" and "60 years old and never operated at this modular scale" — both framings are true; the company only tells the first.
• First-of-a-kind schedule risk. Nuclear FOAK routinely slips years; "first plant online 2026" has been quoted since ~2023 and is now effectively the RELLIS pilot (PWR-5), with commercial units pushed to 2028.

Regulatory findings (required sub-section):

• SEC (EDGAR LR/AAER): none possible — no CIK; private, not an SEC filer. total_sec_findings: 0.
• Litigation — Last Energy is a plaintiff, not a defendant: it co-filed State of Texas v. NRC (W.D. Tex., 6:24-cv-00507, filed Dec 30 2024) challenging the NRC's "utilization facility" rule under the Atomic Energy Act of 1954, arguing the NRC lacks authority over sub-GW reactors. NRC moved to dismiss; the motion was denied as moot Apr 8 2025; plaintiffs expanded (Utah, Louisiana, Florida, Arizona legislature + Deep Fission, Valar Atomics); case stayed repeatedly through 2025-26 and remains unresolved as of this dive. This is a strategic offensive suit, not an enforcement action against the company — but the underlying signal is that its US path depends on winning a novel deregulation argument or routing around the NRC via the DOE pilot pathway.
• Non-SEC enforcement (FTC/DOJ/FDA/etc.): web search surfaced no enforcement action, consent decree, fine, or penalty against Last Energy.
• Conclusion: No material regulatory or legal enforcement findings against the company — verified via SEC EDGAR EFTS (no CIK), web search, and the public docket, as of 2026-07-10. The only litigation is one it initiated.

---

Phase D — Project & stress-test

Lens 11 · IPO-readiness & path-to-tradeable (+private swap for "Forward Projection")

No EPS to project (pre-revenue). The +private question is when does this become tradeable, and on what milestones?

• Stage: growth / late-development. Not pre-IPO.
• IPO-readiness score: 2 / 5. Rationale: real regulatory admissions (UK licensing) and a large nominal order book pull it above a pure seed name, but zero revenue, zero operating units, no crossover fund, no disclosed valuation, and a company that raised only $100M in Dec-2025 all cap it well short of pre-IPO. A microreactor peer set that is public (Oklo, NuScale) sold off 27-30% in H1'26 — the IPO window for pre-power nuclear is cooling, not opening.
• Milestones that would unlock an S-1 / tradeable path (in order):

1. RELLIS PWR-5 reaches criticality (2026) — converts "developer" to "operator." Without this, nothing else matters.
2. First commercial PWR-20 delivering PPA electrons (Wales/Poland; realistically 2027-28, not 2026).
3. A crossover-fund-priced round at a disclosed valuation (the true IPO-proximity signal, currently absent).
4. UK site licence decision (targeted Dec 2027).
5. Project debt financing closed at scale (converting EXIM-style LOIs into drawn facilities).

• Path-to-tradeable estimate: not before 2028-29, and contingent on #1. No Brier forecast logged (per --watchlist rules, and there is no scoreable EPS line).
• Ledger housekeeping: last-energy is absent from research/private-watch.json — recommend adding {beat: energy, stage: growth/late-dev, ipo_readiness: 2, lead_investors: "Astera, Gigafund, Haskell", catalyst: "RELLIS PWR-5 criticality 2026; UK site licence 2027", dossier: "companies/last-energy/deep-dive-2026-07-10.md"}. Not written in this unattended run (wave boundary); flagged for the master session.

Lens 12 · Bull vs Bear

Bull case. Last Energy is the one microreactor developer that removed both schedule-killers — novel fuel and novel physics. On <5% LEU it never touches the HALEU bottleneck that will gate Oklo/X-energy/TerraPower for years. It already did the single hardest Western thing — got a site into UK nuclear licensing (first since 1978) — and has a live DOE-pathway pilot (RELLIS) with pits poured, components built and fuel ordered for a 2026 criticality. If that reactor lights up, Last Energy flips from "PowerPoint pipeline" to "the first Western micro-PWR actually selling electrons," monetizing a $18.9B contracted-power book as a build-own-operate IPP — recurring cash, not one-off machine sales — with a manufacturing learning curve no rival is yet on. In a world where the binding constraint on AI is firm power delivered in 24 months on half an acre with no water, that is a category-defining position.

Bear case (permanent-impairment risks). (1) It has never operated a reactor. FOAK nuclear slips; "2026" has been the target since 2023 and is now a 5 MWe pilot, with commercial units at 2028+. (2) The order book is air until a plant exists — PPAs with SEZs and unnamed data centres are cancellable options, not financed backlog; "$18.9B" is lifetime revenue on zero delivered MWh. (3) The capital model is unproven at fleet scale — $163M of equity funds development; each unit needs project debt (EXIM's $104M LOI was for one Welsh reactor), and the fleet capex for 10,000 units is a number nobody has financed. (4) US-regulatory avoidance — suing the NRC and building abroad reads as inability to clear domestic licensing. (5) No crossover fund and no disclosed valuation — the smart late-stage money hasn't priced it.

Pre-mortem (18 months out, thesis broke): the RELLIS pilot slipped past 2026 (a valve, a fuel-delivery delay, a DOE-authorization snag), the UK licence timeline extended, one or two PPA counterparties quietly walked as their own data-centre plans changed, the next raise was a flat/down bridge with still no crossover fund, and the public microreactor comps (already −30% in H1'26) re-rated the whole category lower — leaving Last Energy a well-narrated developer that still hadn't sold a kilowatt-hour.

Are multiples too high? Unanswerable — no disclosed valuation. But the public proxy set de-rated ~30% in six months while still pre-revenue, which says the market is losing patience with nuclear stories that don't produce power. That is the mood Last Energy must beat.

Contrarian view (what the market refuses to see): amid HALEU-fueled hype names, the boring <5%-LEU PWR is the one with a clean fuel path — the market is over-paying for exotic physics and under-pricing the developer whose only novelty is manufacturing and financing. If any Western microreactor actually ships firm power this decade, the base-rate favorite is the one using the fuel that already exists.

Lens 13 · Devil's Advocate (short-seller)

Dismantling the bull case:

• Where the money is concentrated: in promises. 100% of "revenue" is undelivered PPAs; a single category-shock (data-centre capex pause, AI-power glut, a rival shipping first) evaporates the pipeline's optionality overnight.
• Why the moat is weaker than bulls think: "PWR + off-the-shelf" is not proprietary — NuScale, Rolls-Royce SMR, Holtec, Westinghouse AP300 are all LWR/PWR and several are ahead on licensing (NuScale is NRC-certified). Last Energy's real edge is smallness + own-and-operate, which is a business-model bet, not a technology moat — and business models are copyable.
• Most dangerous competitor bulls underestimate: not Oklo (different fuel/physics) but a credible incumbent — a Westinghouse/Holtec deciding micro-PWR-as-a-service is worth doing, arriving with an operating track record, a real factory, and a balance sheet. Or Radiant, which already has hardware in a DOE test bay and a 20-unit Equinix preorder — further along on demonstrated iron.
• Worst capital-allocation / governance tells: suing your own regulator; refusing to publish a price; a Wikipedia promo-flag; a cap table with no crossover discipline.
• Assumptions that must hold for today's (implied) price: RELLIS criticality on schedule; UK licence by 2027; PPAs convert; project debt materializes at scale; PWR-20 hits its cost target (unknown). All five, roughly on time. That is a long conjunction of low-base-rate events.
• If deployment slips 20-30%: for a pre-revenue developer, a 2-3 year slip isn't a haircut — it's a refinancing event into a colder market, with dilution or a down round the likely outcome.
• Single permanent-impairment scenario (and plausibility ~moderate): the RELLIS pilot suffers a serious technical failure or multi-year slip and a licensed competitor delivers firm power first — Last Energy becomes a stranded developer whose IP (a 60-year-old PWR derivative) has little standalone value.

Lens 14 · Management Questions (ordered by information value)

1. What is the all-in overnight capex per PWR-20 and the target PPA price ($/MWh), and at what cumulative unit number do you hit it? (The one number you won't publish — it decides the entire thesis.)
2. Is the RELLIS PWR-5 still on track for 2026 criticality — and what are the top three things that could push it into 2027?
3. Who on the team has personally licensed, built, or operated a nuclear reactor before, and in what role?
4. How is each PWR-20 project financed — what's the debt/equity per unit, and who lends against a first-of-a-kind fleet with no operating history?
5. Of the ~80 units and $18.9B in PPAs, how many are firm/financed vs cancellable MOUs, and what are the termination rights if you miss delivery dates?
6. What happens to your US business if the NRC lawsuit fails — is the DOE pilot pathway a durable route to commercial US deployment or only to a demonstrator?
7. What is the serial-production factory plan — location, capex, capacity, and when does line one run?
8. How does a permanently sealed, swap-out core work commercially and legally — who takes title to the spent, steel-encased unit, and at what cost?
9. What is the realistic first-commercial-electron date (not pilot) for Wales and Poland, and what's gating it?
10. Why has no tier-1 crossover investor priced a round, and what valuation did the Series C close at?
11. How do you defend against a Westinghouse/Holtec/NuScale deciding to offer micro-PWR-as-a-service with an operating track record?
12. What's the fuel-supply contract behind the Sept-2026 core — single-source or diversified, and priced through the fleet ramp?
13. What insurance/liability structure covers a distributed fleet of 20 MW reactors on third-party industrial sites, and what does it cost per unit?
14. What is current annual cash burn, and how many quarters of runway does the $100M Series C buy through the pilot?
15. At 10,000 units, where does the LEU, the skilled operators, and the ~10M tons of core steel come from — and does that supply chain stay "off-the-shelf" at scale?

---