Kairos Power

Phase A — Understand the business

Lens 1 · Company Overview

Kairos Power is a vertically-integrated advanced-nuclear developer building the KP-FHR — a fluoride-salt-cooled, high-temperature reactor in a small-modular form factor. Founded 2016 in Alameda, California, out of UC Berkeley advanced-reactor research; ~514–550 employees as of Mar 2026. Private, pre-revenue, no ticker.

How it will make money (eventually): it is not a reactor-kit vendor — it intends to develop, build, own, and operate plants and sell electricity, ancillary services, and environmental attributes under long-term PPAs. That is the explicit structure of the Google Master Plant Development Agreement: Kairos sells power, not hardware. So the terminal business model is closer to an IPP (independent power producer) with a proprietary reactor than to an equipment OEM — capital-intensive, offtake-anchored, recurring once plants run.

The technology in plain terms: TRISO fuel — ceramic-coated uranium "poppy-seed" particles (originally an Oak Ridge invention) — packed into graphite pebbles, sitting in a bath of FLiBe (a lithium-fluoride/beryllium-fluoride molten salt first demonstrated at ORNL's Molten Salt Reactor Experiment in the 1960s). The salt runs at near-atmospheric pressure and high temperature, so there's no giant pressure vessel and (the pitch) inherent, passive safety. Fuel is HALEU (uranium enriched 5–20% U-235).

Products / roadmap:

• Hermes 1 — 35 MWthermal, non-power (generates heat, not electricity), Oak Ridge TN, ~$100M project.
• Hermes 2 — two-unit demonstration plant, single reactor uprated 28→50 MWe, first Kairos unit to make electricity; demonstrates the full commercial plant architecture.
• Commercial KP-FHR — plants of twin ~75 MWe reactors (~150 MWe); target: firm power "cost-competitive with natural gas," ~140 MWe cited in some materials.

Main customers: Google (anchor offtaker), TVA (grid counterparty), and — structurally — hyperscaler data centers hungry for 24/7 carbon-free power. Suppliers/partners: DOE (fuel + funding), Los Alamos & Oak Ridge national labs (fuel + testing), Materion (salt). Competitors: Oklo, NuScale, X-energy, TerraPower, plus GE-Hitachi/Rolls-Royce SMR (Lens 7).

Lens 2 · Supply Chain

Map upstream → Kairos → end customer, named nodes:

Fuel chain (the hard part): DOE-sourced HALEU (Jan 2026 contract for Hermes 1 startup) → TRISO particles developed at Kairos's TRISO Development Lab (TDL) and Pebble Development Lab (PDL), Albuquerque → fabricated into pebbles at Los Alamos National Laboratory's Low-Enriched Fuel Fabrication Facility (LEFFF). Oak Ridge National Laboratory runs a 4th, 5-year $27M partnership testing TRISO pebble-fuel lifecycle and reactor tech.

Coolant chain: FLiBe purified at the Molten Salt Purification Plant (MSPP) in Elmore, Ohio, a partnership with Materion Corporation (commissioned 2022) — produced 14 metric tons of FLiBe for the ETU program. Kairos broke ground on its own New Mexico Salt Production Facility in Oct 2024 to internalize supply.

Manufacturing chain: the Albuquerque Manufacturing Development Campus builds reactor vessels in-house — ETU 2 was the "first ASME U2-stamped vessel" built there, from 30+ skid-mounted modules; ETU 3's 14-ft vessel used electron-beam welding with Cambridge Vacuum Engineering and the University of Sheffield AMRC. Construction uses precast concrete panels with 3D-printed sinusoidal joints.

Chokepoints / single-source dependencies:

• HALEU — the single biggest external dependency. Today essentially DOE-allocated (US commercial HALEU enrichment — Centrus — is nascent). Every advanced-reactor developer is fighting for the same scarce HALEU; Kairos's Jan-2026 DOE contract covers Hermes 1 only, not the commercial fleet.
• Beryllium (for FLiBe) — Materion is one of very few Western Be processors; toxic, tightly regulated. Kairos's own salt plant reduces but doesn't eliminate the Be-feedstock dependency.
• TRISO at scale — no one has ever mass-produced TRISO pebbles at commercial-reactor volumes. This is a genuine chokepoint the whole sector shares.
• NRC throughput — the regulator itself is a serial resource (Lens 10).

Verdict on the chain: Kairos is more vertically integrated than any SMR peer (salt, fuel, vessels all in-house-or-lab), which is the deliberate moat play (Lens 3). The chain's fragility is concentrated upstream in HALEU/beryllium, exactly where vertical integration can't reach.

Lens 3 · Competitive Advantages (moats)

1. Regulatory lead — the deepest moat. Kairos holds the first NRC construction permit ever issued for a non-light-water advanced reactor (Hermes 1, Dec 2023), plus a second (Hermes 2, Nov 2024), and 14 NRC-approved topical reports as of Jan 2026 with another under review. Topical reports are reusable licensing building-blocks — every one approved is a de-risked brick competitors haven't laid. This is a years-long, capital-heavy, non-purchasable head start on the single hardest gate in nuclear.
2. Iterative-development process moat. The "Plan-Design-Build-Test" loop — build non-nuclear, reactor-scale ETUs first, iterate hardware in a low-regulation environment, then introduce fuel — compresses the learning cycle that historically took decades. ETU 1 already moved 14 tons of live salt and logged 1,000+ hours of pumped-salt operation. This is a process advantage (SpaceX-style) that's culturally hard to copy, not just a patent.
3. Vertical integration → cost control. In-house salt, fuel, and vessels give schedule and cost control competitors who rely on the supply chain lack — the explicit lever for the claimed 40–50% capital-cost reduction across the first ~7 builds.
4. Anchor-customer validation. The Google master agreement (world's first corporate advanced-reactor purchase) and the TVA PPA (first US utility PPA for a Gen-IV reactor) are commercial proof no private peer except X-energy/Amazon can match.

Bargaining power: weak today, structurally improving. Pre-first-power, Kairos needs Google/DOE more than they need Kairos (many horses in this race). Post-first-power, a working, licensed, cost-competitive Gen-IV reactor with a fuel/salt supply chain it controls would flip that — hyperscalers are desperate for firm clean MW and there are very few credible suppliers.

Lens 4 · Traction & Unit Economics (+private swap — no revenue segments)

No revenue, no segments, no P&L — pre-commercial. Traction is measured in order book + milestones, not dollars:

• Order book: Google Master Plant Development Agreement — 500 MW by 2035, first reactor 2030, ~6–7 reactors (one 50 MW Hermes 2 + six ~75 MW commercial). TVA PPA for up to 50 MW from Hermes 2.
• Unit economics (targeted, unproven): commercial ambition is firm power cost-competitive with natural gas; Laufer frames 40–50% capex reduction as achievable over the first seven reactors via learning-curve + vertical integration. No published $/MWh or $/kW figure — n/a — not disclosed. The honest read: today Hermes is a ~$100M+ science project per unit; the entire equity story is the slope of the cost-down curve, which is 100% unproven at commercial scale.
• Cash-burn signal: ~550 staff + three construction sites + fuel/salt plants ⇒ a heavy, sustained burn funded by DOE milestones + private/debt, not offtake revenue (none until ~2030).

---

Phase B — Measure performance (+private overlay — funding, not earnings)

Lens 5 · Funding & Valuation Trajectory (+private swap for "Earnings Result")

Kairos is a deliberate funding outlier: it has avoided the headline VC/crossover round the rest of the frontier chased, and discloses almost nothing. What's verifiable:

• DOE ARDP Risk-Reduction award (2020): total project value $629M, DOE share $303M, Kairos matching the balance from private sources. Finalized Feb 21, 2024 as a performance-based, fixed-price milestone contract — Kairos is paid only on hitting defined technical milestones.
• HALEU contract (Jan 2026): DOE-sourced HALEU for Hermes 1 — supply-chain funding-in-kind, not cash.
• Private equity: venture backer Terra Talent named; no disclosed round size, no valuation, no crossover-fund (Fidelity/T. Rowe/Coatue) mark — n/a — not disclosed. Total capital "exceeds $600M when government awards are included", but that conflates grant + equity + in-kind.
• Google is an offtaker, not an equity holder — the deal is a PPA/development agreement; no equity stake disclosed. Important: this means no Google balance-sheet backstop the way Amazon's $700M equity underwrites X-energy.

Read: funding quality is real but opaque and milestone-gated. The absence of a priced crossover round is a double-edged signal — it preserves control and avoids down-round risk, but it also means there is no market valuation mark to underwrite an IPO against (Lens 11).

Lens 6 · Founder / Executive Messaging (sentiment trend) (+private swap — interviews, not earnings calls)

Tracking Laufer/Blandford across the public record (Senate testimony, C.O.B. Tuesday, Energy Intelligence, Latitude Media, Marketplace, ANS):

• Consistent core message (stable 2023→2026): "the main challenge with new nuclear is cost"; the answer is iterate cheaply on non-nuclear hardware first. This has not wavered — a good sign of a real thesis, not a pivoting pitch.
• Rising note — urgency (2025→2026): Laufer: "The only thing that has changed is the increased sense of urgency to be ready to scale up sooner" and "I would make the case that we're moving faster than many in the field, if not the fastest". Blandford at Hermes 2 groundbreaking: "working with discipline and urgency to meet the current moment". The AI-data-center demand surge visibly raised the tempo.
• What they've grown careful about: hard commercial dates. Commercial demo framed as "as close to 2030 as possible" — a deliberate hedge, consistent with the Hermes 1 permit-deadline slip (Lens 8/10).
• Tone verdict: confident-but-honest engineers, not promoters. The candor about first-of-a-kind delays (below) is a credibility asset versus the more promotional SPAC-era nuclear names.

Lens 7 · Cap Table & Peer Read-Across (+private swap — comps by stage/mechanism, not P/E)

Pre-revenue nuclear cannot be compared on EV/Sales or P/E — every multiple would be fabricated, so the honest comp table is stage · mechanism · NRC status · offtake · public mark:

Closest mechanism comp: X-energy — the other TRISO-pebble-bed developer. Key divergence: X-energy is helium-cooled (HTGR); Kairos is salt-cooled (FHR). X-energy's April-2026 IPO (>$1.4B total funding incl. Amazon's $700M equity) is the cleanest read-across for what a TRISO-pebble developer is worth at the point it goes public — and it went public before first power, on the story. Read-across for a Kairos valuation: a credible band would sit between Nano Nuclear ($1.2B) and NuScale ($5.3B) given Kairos's stronger regulatory position but earlier funding-disclosure stage — call it ~$2–5B ``. Treat as a rough bracket, not a number to trade on.

Lens 8 · Milestone Catalysts (+private swap — funding/product events, not stock moves)

No stock, so "what moves the story" is the milestone tape. The 5-year event ledger:

• 2020 — DOE ARDP Risk-Reduction award ($629M project / $303M DOE).
• Sep 2021 — Hermes 1 construction-permit application filed.
• Dec 14, 2023 — ⭐ Hermes 1 CP approved — first NRC CP for a non-LWR advanced reactor (landmark de-risking).
• Feb 21, 2024 — DOE $303M milestone contract finalized.
• Jul 2024 — Hermes 1 groundbreaking.
• Oct 14, 2024 — ⭐ Google Master Plant Development Agreement (500 MW / 2035) — world's first corporate advanced-reactor purchase.
• Oct 2024 — NM Salt Production Facility groundbreaking; Nov 2024 — Hermes 2 CP approved.
• May–Jul 2025 — first nuclear concrete, Hermes 1; ETU 3 vessel installed (Jul).
• Aug 2025 — ⭐ Google/TVA PPA — first US utility PPA for a Gen-IV reactor.
• Jan 2026 — DOE HALEU contract finalized (fuel secured for Hermes 1).
• Feb 2026 — ORNL TRISO pebble-lifecycle partnership.
• Mar 2026 — ⚠️ NRC extends Hermes 1 completion deadline 2026→Apr 2029 (FOAK development delay — the one clear negative print).
• Apr 17, 2026 — Hermes 2 groundbreaking.
• Summer 2026 (expected) — ETU 3 completion; ETU 2 startup.

Pattern: the story is driven by regulatory approvals and blue-chip validations (up) and schedule slips (down). The market — when there is one — will react to first criticality of Hermes 1 more than anything else on this list.

---

Phase C — Judge people & books

Lens 9 · Management

• Mike Laufer — CEO, co-founder. UC Berkeley nuclear-engineering pedigree; the public face and cost-thesis author. Archetype: technical founder-operator, SpaceX-style iterate-fast conviction, notably candid about FOAK risk. Testified to the Senate Energy Committee.
• Edward Blandford — CTO, co-founder. Reactor-physics/thermal-hydraulics lead; runs the ETU + Hermes technical program; the "discipline and urgency" voice.
• Per Peterson — co-founder. UC Berkeley professor, one of the foremost FHR academics in the US — the scientific credibility anchor (the KP-FHR is downstream of decades of his DOE-funded work).
• Track record: they have built the thing that has never been built to an unprecedented regulatory milestone (first advanced-reactor CP) and landed Google + DOE + TVA. For a pre-revenue deep-tech, that is a strong, quantified delivery record on milestones (they have delivered nothing commercial yet — the caveat that matters).
• Skin in the game / capital allocation: founder-controlled, deliberately capital-disciplined (milestone-gated DOE money, no disclosed dilutive mega-round). Capital-allocation behavior so far is conservative and control-preserving — a plus for alignment, a question mark for whether they can raise the multi-billions commercial scale-up will need without eventually accepting outside marks.
• Red flags: none obvious — no promotional behavior, no related-party disclosures surfaced. The main governance unknown is precisely the opacity (private, undisclosed cap table).
• Founder vs. professional manager: founder-led, which is right for this iterate-to-first-power stage; the open question is the transition to an operating IPP at fleet scale.

Lens 10 · Forensic Red Flags + Regulatory

Accounting/forensic: n/a — private, unaudited, no public financials. There is no income statement, balance sheet, or cash-flow statement to forensically examine — which is itself the headline caveat: investors get zero audited visibility. The relevant "forensic" risks for a pre-revenue private are (a) burn vs. runway (undisclosed — cannot verify), (b) milestone-recognition on the DOE contract (fixed-price milestone structure is actually investor-protective — DOE pays only on delivery), and (c) contingent liabilities from FOAK construction (undisclosed).

Regulatory findings (required):

• SEC (EDGAR EFTS — LR + AAER): 0 findings — Kairos has no CIK and is not an SEC registrant; no enforcement search possible.
• Non-SEC enforcement (web): search of "Kairos Power" (FTC OR DOJ OR FDA OR CFPB OR "consent decree" OR settlement OR fine OR penalty) surfaced no material enforcement actions.
• 10-K Item 3 (legal proceedings): n/a — private, no 10-K.
• NRC (the regulator that actually matters here): overwhelmingly positive — first advanced-reactor CP (Hermes 1 + Hermes 2), 14 approved topical reports. The one negative: the Mar-2026 request to extend Hermes 1's construction-completion deadline from 31 Dec 2026 to 30 Apr 2029, granted May 2026 — Kairos's own filing attributes it to "development delays associated with the first-of-a-kind nature" of the facility. Not misconduct; it is a hard, sourced signal that FOAK schedule risk is live.
• Summary: No material regulatory or legal findings on the enforcement side — verified via SEC EDGAR EFTS (LR, AAER), web search, and the (non-existent) 10-K Item 3 as of 2026-07-10. The material regulatory risk is forward-looking: the NRC operating license (separate from the construction permit) and first-of-a-kind fuel/salt licensing.

---

Phase D — Project & stress-test

Lens 11 · IPO-Readiness & Path-to-Tradeable (+private swap for "Forward Projection")

No EPS to project — pre-revenue for ~5+ years. The relevant projection is distance-to-tradeable-event.

Readiness scale (1–5, per private-watch.json): Kairos ≈ 2–3 (growth / late-stage) — NOT pre-IPO. Reasoning: the technical/regulatory maturity is late-stage (permits, salt flowing, fuel secured, anchor offtake), but the capital-markets posture is deliberately early — no crossover round, no disclosed valuation, no stated S-1 intent, no secondary-market program disclosed (some accredited-investor exposure exists via Forge/UpMarket secondaries, but that's a thin signal). Kairos has the milestones its public peers (Oklo via SPAC, NuScale, X-energy via IPO) used to go public pre-first-power — and has chosen not to.

Milestones that would unlock an S-1 / tradeable event:

1. Hermes 1 first criticality / first power (~2028) — the single de-risking event; converts "unproven FHR" into "operating Gen-IV reactor." This is the gate.
2. Hermes 2 grid electricity to TVA (~2030) — proves the commercial plant architecture and revenue model.
3. A priced equity round with crossover/strategic marks — none yet; its appearance would be the clearest IPO-proximity tell.
4. Commercial cost-curve evidence — any real $/MWh datapoint proving the natural-gas-competitive claim.

Estimated window: if Kairos follows peers, an IPO is plausible 2028–2030 around first power — but on current signals it is more likely to stay private longer than the SMR cohort, funding the fleet ramp through DOE + strategic + debt until commercial revenue arrives. ``. Do not model this as imminent.

No Brier forecast logged (unattended watchlist run; forecast.ts create skipped per skill + task).

Lens 12 · Bull vs Bear

Bull case. Kairos is the best-positioned Gen-IV developer on the two things that actually gate the sector: licensing and cost-process. It owns the first (and only, alongside its own Hermes 2) advanced-reactor construction permits, has real molten salt pumping in reactor-scale test units, has secured fuel, and has the two most credible offtake validations in the industry (Google, TVA). The iterative, vertically-integrated, non-nuclear-first method is a genuine SpaceX-style process moat that could deliver the 40–50% cost-down that makes nuclear competitive with gas — precisely the firm, 24/7, carbon-free power hyperscalers will pay a premium for through 2035. If Hermes 1 reaches criticality near 2028 and the cost curve bends, Kairos becomes a generational IPP and a top-2 read-across for the entire nuclear-for-AI trade.

Bear case (permanent-impairment risks).

1. First-of-a-kind execution. The permit deadline already slipped 2+ years. FHR + TRISO-pebble + FLiBe at reactor scale has never been operated commercially; salt chemistry, tritium management, and materials corrosion are unforgiving. A serious Hermes 1 setback resets the whole timeline and every downstream PPA.
2. The cost claim is unproven — and it's the whole thesis. "Competitive with natural gas" is an aspiration on a learning curve nobody has walked. If FOAK costs stay stubbornly high (as they did for NuScale, whose flagship UAMPS project died on cost), the offtakes don't pencil.
3. HALEU / fuel-supply scarcity. The commercial fleet needs HALEU that barely exists at commercial volume; a supply squeeze caps the ramp regardless of reactor readiness.

Pre-mortem (18 months out, thesis broke): Hermes 1 hit a salt-loop or materials problem in commissioning, criticality slipped past 2029, a HALEU allocation shortfall appeared, Google quietly re-weighted toward faster gas/solar+storage for near-term data-center load, and — with no priced round and burn rising — Kairos had to accept a down-round or heavy dilution to fund the fleet.

Contrarian view (what the market is refusing to see): because Kairos is private and un-hyped relative to Oklo's meme-stock ascent, the market under-weights that Kairos is arguably the technical leader — it has the permits and the operating salt hardware Oklo doesn't. The mispricing (when it becomes tradeable) could cut bullish, not bearish.

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

Dismantling the bull case:

• Revenue concentration = one customer, one deal, one decade out. The entire commercial story is Google's 500 MW by 2035 and a 50 MW TVA sliver — and Google holds no equity, only PPAs it can renegotiate or de-prioritize if reactors slip while its data-center load can't wait. Concentration doesn't get more extreme than "first revenue ~2030, from one hyperscaler."
• The moat may be thinner than bulls think. A construction permit is permission to build, not a working reactor — NuScale had a design certification and still cratered on economics. The regulatory lead evaporates if the physics-at-cost doesn't work.
• Most dangerous competitor bulls underestimate: X-energy. Same TRISO-pebble fuel, now public with $1B+ raised and Amazon's $700M equity — better capitalized, similar tech maturity, arguably a faster path to fleet financing. And TerraPower (Gates) is also building.
• Capital-allocation / governance: the very opacity bulls read as "discipline" means no audited financials, no valuation mark, no visibility into runway — a short's dream if burn is worse than assumed. There's no way to verify the balance sheet.
• Assumptions that must hold for any "price": (a) Hermes 1 works near 2028; (b) the 40–50% cost-down is real; (c) HALEU is available at fleet scale; (d) Kairos can raise multi-billions without a crushing down-round. If commercial cost lands 20–30% worse than the natural-gas-parity claim, the plants don't clear and the offtakes are renegotiated or void — a plausible, not tail, scenario given nuclear's FOAK history.
• Single permanent-impairment scenario: a serious materials/salt-chemistry failure in Hermes commissioning that proves the FHR concept uneconomic at scale — moderate probability, catastrophic payoff.

Lens 14 · Management Questions (ordered by information value)

1. What is the all-in $/MWh (or $/kW) target for the first commercial KP-FHR, and what is the actual, measured cost of Hermes 1 to date versus your original budget?
2. What is your current cash runway in months, and what total capital do you need to reach first commercial power — and how will you raise it without a dilutive down-round?
3. What is the realistic date for Hermes 1 first criticality, and what is the single most likely technical reason it slips again?
4. Where will commercial-scale HALEU come from for the full Google fleet, and what is your contracted vs. uncontracted fuel position through 2035?
5. What did the 2026→2029 Hermes 1 deadline extension actually cost you in schedule and dollars, and what did you learn that changes the commercial design?
6. Does Google hold any equity or capital commitment beyond the PPAs, and what are the price and termination terms if reactors are late?
7. What is your latest priced valuation and cap table, and why have you avoided a crossover round the rest of the sector took?
8. On the 40–50% cost-down curve — which specific line items deliver it, and what's your evidence the learning rate is real and not aspirational?
9. What is your plan and timeline for the NRC operating license (vs. the construction permit you hold), and what's the critical-path licensing risk?
10. How do you manage tritium, salt corrosion, and materials degradation over a commercial reactor's life — and what has the ETU/Hermes data shown so far?
11. Build vs. own-and-operate: are you committed to being an IPP at fleet scale, and how do you finance owning gigawatts of plant?
12. What is your IPO / liquidity intent and timeline, and which milestone triggers it?
13. How do you defend against X-energy (same fuel, now public and Amazon-backed) and TerraPower on both technology and capital access?
14. What is your TRISO mass-production plan beyond the Los Alamos LEFFF pilot, and what volume/yield have you demonstrated?
15. Beyond Google/TVA, what does the next tranche of the order book look like, and how many binding MW are under contract today?

---