Helion Energy

Phase A — Understand the business

Lens 1 · Company Overview

Helion Energy builds pulsed magneto-inertial fusion machines and intends to sell the electricity they produce. Founded 2013 in Everett, Washington by David Kirtley (PhD aerospace/plasma, Michigan; ex-AFRL, ex-NASA Glenn, ex-MSNW) with John Slough, Chris Pihl, and George Votroubek — the team spun out of MSNW, a Seattle plasma-propulsion shop. Headcount is 500+.

The business model is unusual for a deep-tech startup: it is a power-generation company, not an equipment vendor. Helion does not plan to license reactors or sell hardware to utilities. It builds, owns, and operates the machine and sells the output under long-dated power-purchase agreements (PPAs). The commercial engine is:

• Microsoft PPA (May 2023) — the world's first fusion PPA. Helion commits to bring a generator online by 2028 and reach 50 MW+ within a one-year ramp, delivering to a Microsoft data center in Central Washington. Constellation Energy is the power marketer / transmission manager. Critically, this is a real PPA with financial penalties if Helion fails to deliver — CEO Kirtley has confirmed the penalty clauses on record. This is the single most important fact about the company: a hyperscaler wrote penalty clauses against a 2028 fusion date.
• Nucor collaboration (Oct 2023) — a 500 MWe plant at a Nucor steel facility, targeting operations ~2030. Nucor is the largest steel producer/recycler in North America and also a Series-F investor.
• OpenAI (2024, non-binding option) — an agreement giving OpenAI "the opportunity" to buy Helion power in the future; no power has ever been delivered. (See Lens 9/13 — this is the Altman conflict.)

Pricing claim: Helion says its Microsoft/customer contracts guarantee $0.01/kWh ($10/MWh), without assuming mass-production scale economies, carbon credits, or subsidies. If real, that undercuts essentially all grid power. It is also unverified — there is no public disclosure of the actual $/MWh in the Microsoft contract.

Plain-terms summary: Helion is a bet that a 60-foot pulsed-plasma tube with no steam turbine can be manufactured cheaply, run reliably at high pulse rates, and sold as baseload power to AI data centers and heavy industry — and that it can do so years before any tokamak competitor. The revenue is entirely forward: there is no product and no revenue today.

Lens 2 · Supply Chain

The value chain, named end to end:

Upstream inputs → Helion → end customer:

• Fuel — deuterium (cheap, abundant, extracted from seawater) and helium-3 (the commercial-target fuel; see chokepoint below). Interim testing fuel is tritium, which Helion is licensed to possess (first US fusion firm licensed to do so).
• High-power pulsed-power electronics — capacitors, semiconductor switches (IGBTs/solid-state), and the direct-conversion capture electronics are the core of Helion's cost thesis. Helion designs and builds much of this in-house; the pulsed-power supply chain (capacitor banks, switching) is the real industrial dependency and overlaps with the defense/laser/grid-scale-inverter supply base. Specific vendors are undisclosed.
• Superconducting / high-field magnets — the FRC accelerator and compression stages need large fast magnets. Unlike Commonwealth Fusion (whose whole thesis is HTS REBCO tape), Helion's design leans on pulsed magnetic compression rather than steady-state HTS — reducing (not eliminating) exposure to the HTS-tape supply bottleneck that constrains the tokamak camp.
• Manufacturing — Everett/Malaga, WA facilities; Helion has explicitly framed manufacturability of the machine at volume as its differentiator (build the 8th, 9th, 10th generation faster/cheaper). This is vertical integration by design.

Chokepoint 1 — helium-3. The commercial D-³He fuel cycle depends on an isotope that is exceedingly scarce on Earth with no natural terrestrial source at fleet scale; historically sourced from tritium decay in nuclear-weapons stockpiles. Helion's answer is a closed-loop cycle: D-D side reactions produce ³He directly and tritium (which β-decays to ³He, 12.3-yr half-life); Helion captures, separates, and recycles the isotopes on-site. This is elegant and unproven at scale — it is a second physics/chemistry problem stacked on top of the first.

Chokepoint 2 — end customer concentration. Demand today is effectively two names: Microsoft and Nucor (plus the optional OpenAI overhang). The buyer side is a handful of hyperscalers/industrials, so customer concentration is extreme — but so is the willingness-to-pay, because AI data-center power demand is the tailwind (Lens 8).

Names or it didn't happen: Microsoft (offtaker), Constellation Energy (marketer/transmission), Nucor (offtaker + investor), OpenAI (optional offtaker), and on the input side the pulsed-power/capacitor/switch base (vendors undisclosed). The chain is short and concentrated, which is both the risk and the reason a 2028 date is even conceivable.

Lens 3 · Competitive Advantages (moats)

Helion's putative moats, ranked by durability:

1. No steam cycle → structural cost advantage (if it works). Helion's direct-energy-conversion approach recaptures electricity inductively (Faraday's law: expanding fusion plasma pushes on the magnetic field, inducing current) with no turbine, no cooling tower, no steam. If direct conversion hits anticipated efficiencies, the balance-of-plant cost floor collapses relative to every thermal-cycle competitor (tokamaks, most fission). This is the whole bull moat — and its maturity is the whole bear risk (direct conversion is far less proven than thermal conversion; cost uncertainty is "much larger" per independent analysis).
2. Manufacturability / iteration speed. Seven generations of machines built (Polaris is Gen-7); the design philosophy optimizes for cheap, repeatable, mass-manufacturable pulsed units rather than one giant ITER-scale build. In a fleet-deployment world this is a scale/process moat.
3. Regulatory first-mover. Helion is the first company to receive US regulatory approval to possess and use tritium for fusion demonstration, and in June 2026 became the first to clear Washington-state licenses for handling radioactive materials for a fusion power plant. Under the new fusion-specific regime (Lens 10), being first through the door is a real, if temporary, lead.
4. The Microsoft/Constellation/Nucor commercial relationships + Altman/SoftBank/Thrive capital base. Access to penalty-backed offtake and to effectively unlimited late-stage capital ($15.5B round) is itself a moat against smaller competitors.

Bargaining power. Weak today, potentially strong later. Helion needs Microsoft (a marquee validating customer) more than Microsoft needs Helion (Microsoft has many power options). But the direction of travel — AI power scarcity — shifts leverage to whoever can actually deliver firm zero-carbon baseload. Over suppliers Helion has ordinary-to-strong power (it designs its own core electronics; commodity inputs).

The honest moat verdict: the moat is a physics bet, not yet a business moat. Everything durable (cost floor, fleet manufacturability) is contingent on the core reaction working at net-positive electricity — which has not been demonstrated (Lens 5).

Lens 4 · Segments

n/a — private, pre-revenue, not disclosed. There are no revenue segments: Helion has zero commercial revenue. The only "segment" structure is the project pipeline, which belongs in Lens 5 (funding) and Lens 11 (path-to-tradeable):

All revenue is forward and contingent on Orion delivering.

---

Phase B — Measure performance

(+private overlay: Lens 5 → Funding & valuation trajectory; Lens 7 → Cap table & secondary marks; plus a Traction/unit-economics note.)

Lens 5 · Funding & Valuation Trajectory (overlay)

Helion is one of the best-capitalized private deep-tech companies in the world. The trajectory, all `` and unaudited:

Total raised: >$1.5B in equity to date.

The signal in the marks: the valuation went $5.425B → $15.5B, a 2.8x step-up in ~17 months. That re-rate was driven by (a) the Feb-2026 Polaris D-T milestone and (b) the AI-power-demand narrative, not by net-electricity proof (which still hasn't happened). Helion is now, by a wide margin, the most expensive fusion company on Earth — roughly 2.5–3x the ~$5–6B implied value of Commonwealth Fusion and ~2.5x the ~$6B TAE/Trump-Media SPAC value (Lens 7). The burn signal: a ~$465M raise against a 2028 first-plant plus a 500 MW second plant implies Orion + Nucor construction is capital-hungry; the raise cadence (roughly annual, escalating) is consistent with heavy pre-revenue burn — exact burn undisclosed.

Traction / unit economics (overlay note): revenue run-rate = $0. The only "unit economic" is the claimed $0.01/kWh contract floor, which is (i) unverified and (ii) explicitly caveated by analysts as unknown against the actual Microsoft $/MWh, meaning even a technical success could be a cash failure if opex > contract price. n/a — not disclosed on margins, ARR, or burn multiple.

Lens 6 · Founder / Public Sentiment Trend (overlay: founder interviews, not earnings calls)

No earnings calls exist. Substituting the public-communication trend from Kirtley interviews and Helion's newsroom, tracked over ~2023→2026:

• Consistent message: manufacturability + direct conversion + "fusion this decade." Kirtley reframes his career as "from starships to sustainable star power".
• What shifted (the tell): the language moved from "net electricity from Polaris" (the 2021-era promise, targeted 2024) to "industry-first milestones" / "measurable D-T fusion" / "150M °C" (the 2026 framing). The company stopped leading with the one metric that matters (net electricity) and started leading with intermediate physics milestones. That is a classic pre-proof pivot in messaging — impressive-sounding sub-milestones substituting for the gating result.
• New emphasis: regulatory firsts (tritium license, WA licenses) and the Microsoft 2028 date — i.e. commercial/regulatory momentum foregrounded while the core physics proof is pending.

Sentiment is bullish and disciplined on Helion's own channels; independent physicist sentiment is skeptical (Lens 13).

Lens 7 · Cap Table, Syndicate Quality & Fusion Comps (overlay)

Syndicate quality — the IPO-proximity read. Helion's cap table is now tier-1 / crossover-heavy, which is the be-early tell that a private is approaching tradeable:

• Crossover / growth funds: Thrive Capital (Series-G lead — Josh Kushner's fund, a late-stage/IPO-proximate investor), Lightspeed, Lux Capital, SoftBank Vision Fund 2, Peak XV (ex-Sequoia India). Thrive leading at $15.5B is a meaningful pre-IPO signal.
• Strategics / offtakers on the cap table: Nucor (also the 500 MW customer), plus Bill Ford personally.
• Founder/insider whales: Sam Altman owns ~one-third of Helion, a stake reported at ~$1.65B as of late 2025 (he has invested ≥$375M). Altman is the largest individual holder.
• Notable absence vs peers: no classic mutual-fund (Fidelity/T. Rowe) markup disclosed, so no public secondary/markdown mark to triangulate — valuation is primary-round-set only.

Fusion peer comparison table — private-market valuations, ``; there are no P/E, EV/EBITDA, or ROE multiples because none of these companies has revenue. Comparing by capital raised and private valuation is the only honest frame:

Read: Helion trades at a ~2.5–3x premium to every peer on a fraction-to-parity of the capital raised. The premium is priced entirely on (a) the Microsoft penalty-backed 2028 date and (b) the direct-conversion cost story. CFS has raised the most and has arguably the most physics-validated path (HTS magnets are proven; net-gain target 2027). Helion is the highest-valued and least-de-risked-on-core-physics of the three — that is the central valuation tension. Multiples: n/a — no revenue at any peer.

Lens 8 · Valuation / Milestone Catalysts (the "what moves the mark") (overlay: private → funding & milestone events, not stock)

Helion has no stock, so the relevant "catalysts" are the events that re-rate the private mark or make/break the thesis:

• Positive re-raters (historical): Microsoft PPA (May 2023) → first fusion PPA, established commercial credibility; Nucor 500MW (Oct 2023); Polaris D-T fusion + 150M °C (Feb 2026) → drove the 2.8x re-rate; Orion groundbreaking (Jul 2025); WA state radioactive-materials licenses (Jun 2026); $465M Series G at $15.5B (Jun 2026).
• The macro tailwind that made all of it possible: the AI data-center power crunch. Hyperscalers are structurally short firm, zero-carbon, 24/7 power; that scarcity is what let a fusion startup sign penalty-backed offtake and triple its valuation pre-revenue.
• The gating catalyst that HASN'T fired (the whole thesis): net electricity from Polaris. Targeted 2024, reset to 2025, missed both years with no replacement date announced as of Dec 2025. This is the single event that would either confirm or break the bull case.

Pattern the "market" (private investors) reacts to: funding/valuation steps have tracked commercial and intermediate-physics milestones — not the net-electricity proof. The market is currently paying for the promise of the gating result before the result exists. That is the definition of a milestone-risk-loaded mark.

---

Phase C — Judge people & books

Lens 9 · Management

• David Kirtley — Founder & CEO. PhD aerospace engineering (Michigan Plasmadynamics Lab); AFRL research scientist (led a coaxial-FRC thruster design); NASA Glenn Lewis Fellow; PI/fusion-lead at MSNW for nine years before founding Helion in 2013. Track record: deep, genuine plasma-physics and pulsed-power pedigree; he has actually built seven machine generations and raised >$1.5B — but he has not yet delivered the one thing he has promised for a decade (net electricity). Founder-technologist archetype, not a professional operator. For a pre-proof deep-tech company that is the right archetype (you want the physicist who understands the machine), but it also means the missed-deadline pattern sits directly on his credibility.
• Skin in the game: founder-led, private; founding team still central. Insider ownership is concentrated in founders + Altman.
• Capital allocation: the relevant history is fundraising and building, not buybacks/M&A. He has consistently raised at escalating valuations and plowed it into machines and now plant construction — appropriate for the stage. No value-destroying capital moves evident.
• Red flag — the Altman relationship (material). Sam Altman chaired Helion's board and is its largest investor (~1/3, ~$1.65B). OpenAI signed a 2024 option to buy Helion power. This created a documented conflict of interest: Altman stepped down from Helion's board as OpenAI explored a power partnership; the US House Oversight Committee requested OpenAI's conflict documents (May 8 letter, citing Helion by name); attorneys general from six states (FL, MT, NE, IA, WV, LA) asked the SEC to scrutinize Altman's conflicts; and the relationship surfaced in the Musk v. OpenAI trial, where Altman testified he had "always been recused". This is a governance overhang on Helion even though Helion itself is not accused of wrongdoing — the company's single largest backer and most valuable customer relationship (OpenAI) sit at the center of a live conflict-of-interest and political-scrutiny story.

Lens 10 · Forensic Red Flags + Regulatory

Forensic accounting: n/a — private, no audited financials. There is no income statement, balance sheet, or cash-flow statement to forensically examine; no SBC, goodwill, or receivables to test. The equivalent forensic questions for a private pre-revenue fusion company are:

• The $0.01/kWh claim is the number to distrust. It is asserted without scale economies/subsidies and with no disclosed contract $/MWh to check it against — treat as marketing until a contract price is disclosed.
• Milestone-date reliability is the "earnings quality" proxy — and it is poor: net electricity targeted 2024 → 2025 → missed both, no new date. A company that has repeatedly reset its gating milestone while raising at rising valuations is the pre-revenue analog of a company guiding down while the multiple expands.
• Burn / runway undisclosed — a ~$465M raise for a 2028 plant plus a 500MW plant is not obviously sufficient to first-power; expect further raises (dilution) or debt/project finance.

Regulatory findings (required sub-section).

• SEC (EDGAR EFTS — LR + AAER): zero findings. Helion has no CIK, is private, and is not required to file; no EDGAR enforcement search is possible.
• Non-SEC enforcement: web search for "Helion Energy" (FTC OR DOJ OR FDA OR CFPB OR consent decree OR settlement OR fine OR penalty) returns no enforcement action against Helion Energy the company. The only adjacent legal exposure is the Altman/OpenAI conflict drawing House Oversight and six-state-AG→SEC scrutiny — directed at Altman/OpenAI governance, not at Helion. Note this is reputational/governance-adjacent, not a Helion enforcement matter.
• Item 3 (Legal Proceedings): n/a — no 10-K exists (private).
• Regulatory regime (favorable — a genuine tailwind): The ADVANCE Act (Jul 2024) amended the Atomic Energy Act so fusion machines are regulated as byproduct material under 10 CFR Part 30 (the particle-accelerator/materials framework), not as fission reactors under Part 50. The NRC codified the split of fusion from fission regulation; a Federal Register "Regulatory Framework for Fusion Machines" proposed rule advanced (published Feb 26, 2026). Part-30 licensing is dramatically lighter than Part-50 reactor licensing. Helion is the first mover through this lighter regime — first tritium-possession approval, first WA plant licenses. This materially de-risks the regulatory path to 2028 even as the physics path remains open.
• Verdict: No material regulatory or legal findings against Helion — verified via SEC EDGAR EFTS (LR/AAER = 0), web search (no enforcement), and the absence of any filing regime, as of 2026-07-07. Governance overhang exists via the Altman/OpenAI conflict but does not name Helion as a respondent.

---

Phase D — Project & stress-test

Lens 11 · IPO-Readiness & Path-to-Tradeable (overlay — replaces EPS forecast)

No EPS. For a +private name the projection lens is: when does this become tradeable, and what unlocks it?

private-watch.json grounding: stage: late, ipo_readiness: 3/5, catalyst: Microsoft PPA (2028). My independent read confirms late but argues 3/5 is if anything generous on an equities-IPO basis — Helion has tier-1/crossover backing (S-1-capable syndicate ✔) and enormous scale ($15.5B ✔), but it has no revenue and no proven core product, so a traditional IPO is implausible until at least net-electricity proof. The realistic path-to-tradeable, in order of likelihood:

1. Stay private and keep raising (base case through 2027) — the AI-power narrative + Thrive/SoftBank/Altman capital means Helion can fund Orion without public markets. No forced liquidity event.
2. SPAC / direct listing on a milestone (plausible 2027–2028) — TAE just took the Trump-Media SPAC route at ~$6B; a Polaris net-electricity result or a successful Orion first-power would be a natural SPAC/IPO trigger for Helion at a far larger number. Fusion has a live public-market appetite (Oklo, NuScale trade publicly).
3. Traditional IPO (2029+) — only realistic after Orion actually delivers power to Microsoft under the PPA, converting the story from physics-bet to revenue.

The milestones that unlock an S-1 (be-early payoff):

• Gating: Polaris demonstrates net electricity (the missing proof). Without this, any listing is a pre-revenue physics bet.
• De-risking: Orion first plasma / first power on the Malaga site (targeted 2028).
• Revenue: first MWh delivered to Microsoft under the PPA (2028 target; one-year ramp to 50MW).

Estimated tradeable window: 2027 (opportunistic SPAC on a physics milestone) to 2029+ (true IPO on delivered revenue). ``

(Per wave boundary, the private-watch.json dossier field is NOT written back by this run.)

Lens 12 · Bull vs Bear

Bull case. Helion is the only fusion company that has converted a physics program into a penalty-backed commercial contract with a hyperscaler — Microsoft bet money on a 2028 date. Its direct-conversion, no-steam-cycle architecture is the one design in fusion with a credible path to a radically lower cost floor (no turbine, no cooling tower), and if direct conversion works at anticipated efficiency, Helion doesn't just win fusion — it undercuts fission and gas. The AI data-center power crunch is a once-in-a-generation demand tailwind that a firm, zero-carbon, 24/7 source is perfectly positioned for, and the new light-touch Part-30 regulatory regime (which Helion is first through) removes the multi-year fission-licensing barrier. Capital is effectively unlimited (Thrive/SoftBank/Altman at $15.5B), and the Feb-2026 D-T fusion at 150M °C proves the machine reaches thermonuclear conditions. Every intermediate milestone has landed.

Bear case (2–3 permanent-impairment risks).

1. The core physics may simply not close. Independent physicists (ex-PPPL's Daniel Jassby: "voodoo fusion") and a 2018 MITRE/JASON/ARPA-E report flag the exact failure mode — Helion's fire-and-compress approach risks catastrophic flux loss / plasma instability before fusion does the needed work; achieving high compression + plasma stability simultaneously is unproven. If net electricity is physically unreachable in this architecture, the company is impaired regardless of capital.
2. The helium-3 fuel cycle is a second unsolved problem. Commercial D-³He needs ³He that doesn't exist at scale; the closed-loop breed-and-recycle scheme is unproven, and D-³He needs ~200M °C (Polaris is at 150M).
3. Even a technical win can be a cash loss. No disclosed Microsoft $/MWh; if opex exceeds contract price, Helion burns equity even while delivering power.

Pre-mortem (it's Jan 2028, the thesis broke). Polaris ran D-T beautifully but never crossed net-electricity; Orion's first-plasma slipped to 2029; Helion paid Microsoft the PPA penalty and quietly reset the date a third time. The $15.5B mark — set on the promise of net electricity — compresses hard on a down-round as crossover funds mark to the missed milestone, and the "most expensive fusion company" becomes the cautionary tale of pre-revenue AI-power euphoria.

Are the multiples too high? There are no multiples (no revenue), but $15.5B is ~2.5–3x every better-capitalized or better-physics-validated peer on a thesis whose gating proof has slipped twice. On a risk-adjusted basis the mark is rich.

Contrarian view (what the market refuses to see). The market is pricing Helion as if the Microsoft date de-risks the physics. It does not — Microsoft de-risked its own optionality (cheap penalty vs. huge upside), not Helion's plasma stability. The crowd is paying a physics-proof premium for a physics-proof that hasn't happened.

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

If Helion were public, here is the short:

• Structural break: the entire enterprise rests on one undemonstrated result — net electricity from a pulsed FRC — that has been promised since 2021 and missed in 2024 and 2025 with no new date. A decade-old "5–10 years away" pattern (Jassby's critique) is the base rate, and Helion fits it.
• Revenue concentration: ~all future revenue is Microsoft + Nucor. If either walks (Microsoft's penalty is cheap relative to its optionality; it can source power elsewhere), the commercial thesis thins to nothing.
• The moat may be weaker than bulls think: direct energy conversion — the sole cost moat — is the least mature part of the stack; its cost uncertainty is explicitly "much larger" than thermal. Meanwhile CFS has proven HTS magnets and a 2027 net-gain target on more capital — the "most dangerous competitor bulls underestimate" is CFS reaching net-gain first and reframing Helion's aggressive date as hype.
• Governance/incentives: the largest backer (Altman, ~$1.65B / ~1/3) is enmeshed in a live conflict-of-interest drawing Congress and six state AGs; the marquee customer (OpenAI) relationship is the conflict. Related-party optics are as bad as they get for a pre-IPO name.
• Valuation: $15.5B, up 2.8x in 17 months, on zero revenue and a twice-missed gating milestone. If net-electricity proof slips again or Orion's 2028 date moves, a down-round is the base case. A 20–30% "growth disappointment" isn't the risk — a binary physics miss that takes the valuation down 50–80% is.
• The single scenario that permanently impairs the business: Polaris demonstrates that Helion's architecture cannot simultaneously hit the compression and stability needed for net electricity — i.e. the JASON risk realizes. Plausibility: genuinely uncertain, non-trivial — credible physicists put real weight on it; this is not a strawman.

Lens 14 · Management Questions (ordered by information value)

1. Has Polaris produced net electricity — measured energy out of the machine exceeding energy in — yes or no, and if not, what is the specific, dated target now? (This is the entire thesis; everything else is secondary.)
2. What is the current best-measured energy-in vs. energy-out ratio on Polaris under D-T, and what is the gap to net-electricity break-even?
3. On the JASON/MITRE concern: what data do you now have that you can simultaneously achieve the required compression and plasma stability without catastrophic flux loss?
4. What is the actual contracted $/MWh in the Microsoft PPA, and at what plant cost does Orion become cash-flow positive rather than merely delivering power?
5. What is the current cash runway, and how many more dollars to Orion first-power at 50 MW — is the Series G sufficient or is another raise required before 2028?
6. What are the specific penalty terms in the Microsoft PPA if 2028 is missed, and what is your internal probability of hitting the date?
7. The D-³He commercial fuel needs ~200M °C and a working closed-loop ³He breeding/recycling system — what have you demonstrated on ³He production and separation at any scale?
8. Why did net-electricity slip from 2024 to 2025 to no-date, and what specifically was underestimated each time?
9. Orion is being built before Polaris has proven net electricity — what is the contingency if Polaris does not reach net electricity before Orion needs to power on?
10. How much of the $0.01/kWh claim survives once you include capital recovery on the plant, and does it assume the direct-conversion efficiency you have actually measured or a target?
11. What is the direct-energy-conversion efficiency you have measured on Polaris, versus the efficiency your cost model assumes?
12. How exposed is your build schedule to the pulsed-power / capacitor / switch supply chain, and are any of those single-sourced?
13. With CFS targeting SPARC net-gain in 2027, how do you frame Helion's differentiation if a tokamak reaches scientific net-gain before your net-electricity demo?
14. Given the Altman conflict-of-interest scrutiny, how is Helion's governance insulated, and does the OpenAI power option survive the recusal / political attention?
15. What does a realistic liquidity path look like for existing investors — stay-private-and-raise, SPAC on a milestone, or IPO on delivered revenue — and on what timeline?

---