Fusion Approach Comparison
| Approach | Confinement | Typical Fuel | Achieved T (MC) | Ignition T (MC) | Confinement Time | Scale-up Challenge | Leading Examples |
|---|---|---|---|---|---|---|---|
| Tokamak | magnetic | D-T | 150 | 150 | 1 to 10+ | Plasma disruptions can damage first wall; large size historically required (ITER 6.2 m major radius); HTS magnets unlock smaller scale (SPARC 1.85 m). |
|
| Stellarator | magnetic | D-T | 100 | 150 | 30+ minutes demonstrated at W7-X | 3D-shaped magnetic coils require extreme manufacturing precision; no plasma current means disruption-free steady-state but engineering is harder. |
|
| Spherical Tokamak | magnetic | D-T | 100 | 150 | 0.1 to 1 | Higher plasma beta in compact geometry but center column engineering and first-wall heat flux are severe. |
|
| Field-Reversed Configuration | magnetic | D-He3 (Helion), p-B11 (TAE target) | 75 | 500 (D-He3), 1500 (p-B11) | milliseconds (pulsed) | Pulsed-merger approach (Helion) needs high-repetition-rate switching; beam-heated approach (TAE) struggles to reach p-B11 temperatures. |
|
| Z-Pinch | magnetic | D-T | 37 | 150 | microseconds; seconds-long with sheared-flow stabilization | Classical Z-pinch is unstable; sheared-flow stabilization is novel and not yet proven at fusion-relevant conditions. |
|
| Inertial Confinement Fusion (direct drive) | inertial | D-T | 100 | 100 | nanoseconds | Drivers must run at ~10 Hz for commercial output; NIF runs 1 shot per day. Target fabrication at scale unsolved. |
|
| Inertial Confinement Fusion (indirect / X-ray drive) | inertial | D-T | 100 | 100 | nanoseconds | Hohlraum target conversion efficiency loss; Q_wall-plug at NIF is ~0.01 even though Q_laser reached 1.5 in Dec 2022. |
|
| Magnetized Target Fusion | magneto-inertial | D-T | 10 | 150 | microseconds | Liquid-metal compression liner adds engineering complexity; reaction-vessel survivability unproven. |
|
| Hybrid / Pulsed-Power ICF / Other | magneto-inertial | D-T (Pacific), p-B11 (HB11) | n/a | 150 to 1500 | nanoseconds to microseconds | Each hybrid is bespoke. Pacific Fusion uses pulsed-power Z-driven ICF (Sandia heritage); HB11 fires petawatt laser at p-B11 targets; Avalanche uses electrostatic orbitron; classifications overlap with FRC, Z-pinch, and ICF. |
|
Nine civilian-power fusion confinement approaches by physics family. Color codes: magnetic (continuous field), inertial (laser or pulsed-power compression), magneto-inertial (hybrid). Plasma temperatures in millions of degrees C; achieved temperatures reflect highest publicly-reported value. Helion is logged as FRC but operates as merged-FRC magneto-inertial; Pacific Fusion is logged hybrid (pulsed-power Z-pinch ICF). Propulsion-focused fusion (Pulsar, Helicity) excluded. Source: FIA Global Fusion Industry Report 2024, ITER Organization, Nuclear Fusion journal review issues. Accessed 2026-05-15.
Fusion Fuel Cycles
Tritium supply chokepoint. Global commercial tritium supply is roughly 3 kg per year, almost entirely from CANDU heavy-water reactors at Ontario Power Generation Darlington. Commercial D-T fusion at scale requires self-sustaining tritium breeding (TBR greater than 1.05) from lithium blankets, never demonstrated at burning-plasma conditions. This is the single largest D-T pathway engineering risk.
| Fuel Cycle | Ignition T (MC) | Tritium | Neutron Output | Fuel Availability | Companies Pursuing | Notes |
|---|---|---|---|---|---|---|
D-T Deuterium-Tritium | 150 | required | high | Deuterium abundant in seawater (33 g per m3). Tritium is rare: ~3 kg per year of commercial supply globally, primarily from CANDU heavy-water reactors (Ontario Power Generation). Commercial fusion at scale requires self-sustaining tritium breeding from lithium blankets, never demonstrated at scale. | ITER, NIF (LLNL), JET, EAST, +15 more | Default fuel for public programs and most private developers because of lowest ignition temperature. Tritium supply chokepoint is the single largest commercial-fusion bottleneck. |
D-D Deuterium-Deuterium | 400 | not required | moderate | Deuterium abundant in seawater. Eliminates tritium supply dependency. | Helion Energy (intermediate step toward D-He3) | Used by Helion as an intermediate fuel cycle to breed He3 onboard for the D-He3 reaction. Not pursued as a standalone commercial fuel by any major program. |
D-He3 Deuterium-Helium-3 | 500 | not required | low | He3 is rare on Earth (defense-decay stockpile + heavy-water reactor byproduct, kg-per-year scale). Lunar regolith concept exists but is decades from feasibility. Helion plans to breed He3 onboard from D-D side reactions. | Helion Energy | Helion's stated fuel target. Reaches usable temperatures at ~500 M-degrees C. Helion harvests He3 from the D-D step. The Microsoft 50 MW 2028 PPA is based on this fuel cycle. |
p-B11 Proton-Boron-11 | 1,500 | not required | near-zero | Both fuels abundant. Proton from hydrogen, boron-11 from natural boron (80 percent of natural B). | TAE Technologies, HB11 Energy, Helion Energy (long-term) | Aneutronic fuel cycle. Eliminates tritium supply chokepoint, neutron-activated structure, and most radioactive waste. Required temperatures are 5-10x harder than D-T; no demonstration of net energy at p-B11 conditions to date. |
p-Li6 Proton-Lithium-6 | 1,000 | not required | low | Lithium-6 requires isotopic enrichment; natural Li is 7.5 percent Li-6. | none | Of academic interest as an aneutronic-adjacent fuel cycle. No private company is targeting p-Li6 as the primary commercial fuel as of data cutoff. |
Five fusion fuel cycles spanning the public-program default (D-T) to aneutronic targets (D-He3, p-B11) and the academic alternative (p-Li6). Required temperatures are textbook-standard order-of-magnitude figures; aneutronic fuels are routinely cited as 5-10x harder than D-T. Source: IAEA tritium handling reports, FIA Global Fusion Industry Report 2024, Helion Energy technology page, TAE Technologies, HB11 Energy, Ontario Power Generation Darlington Tritium Removal Facility, Nuclear Fusion journal. Accessed 2026-05-15.
Cross-Cutting Enabling Technologies
Civilian power scope: ten cross-cutting enabling technologies on which any commercial fusion plant depends. Scientific Q (fusion energy out / heating energy in) is the published milestone; engineering Q wall-plug (electricity out / electricity in at the wall) is the harder, less-cited gating metric. Headline gap: REBCO tape, tritium breeding TBR greater than 1, and 10 Hz MJ-class lasers are all unsolved at commercial duty.
Headline enabler
High-Temperature Superconducting (HTS) Magnets
REBCO (rare-earth barium copper oxide) tape conductors operating at 20 K allow compact tokamaks at 20+ tesla. Key enabler for SPARC, ARC, ST40, stellarator commercial scale.
Primary players
Commonwealth Fusion Systems, Tokamak Energy, Type One Energy, Realta Fusion, Proxima Fusion, Energy Singularity
Suppliers / OEMs
Faraday Factory, Bruker, Sumitomo Electric, Fujikura, SuperPower (Furukawa), Shanghai Superconductor
Critical gap
REBCO tape global production capacity is a single-digit-tonnes-per-year industry; CFS plans demand multi-tonne-per-year supply by SPARC commissioning.
Tritium Breeding Blankets
in designLithium-bearing blankets (Li-Pb eutectic or Li-ceramic pebble bed) surround the plasma to breed tritium via neutron capture on Li-6. Demonstrated at lab scale; never at commercial Tritium Breeding Ratio > 1.
Plasma-Facing Materials
in developmentFirst-wall materials must survive 10+ MW per m2 heat flux, 14 MeV neutron bombardment, and tritium retention.
High-Power Lasers (ICF)
NIF beam energy proven; high-rep-rate at...NIF runs 192 neodymium-glass beams at 1.8 MJ per shot, 1 shot per day. Commercial ICF requires 10 Hz repetition rate (864 thousand shots per day), 100x energy efficiency gain.
Pulsed-Power Drivers
operationalZ-machine class drivers compress targets via current pulses. Sandia Z is the world reference; Pacific Fusion plans 60 MA driver.
Plasma ML and AI Control
demonstratedReal-time disruption avoidance, plasma-shape optimization, and surrogate-model design. DeepMind + EPFL TCV tokamak demonstration Feb 2022 (Nature paper).
Heat Extraction at Greater Than 500 Degrees C
no fusion prototypeHigh-temperature thermal cycle (helium Brayton, supercritical CO2, molten salt) to convert fusion-produced heat to electricity at greater than 40 percent thermal efficiency.
Vacuum Vessel Manufacturing
operationalLarge-diameter, precision-welded vacuum vessels for tokamaks and stellarators. ITER cryostat is the world's largest stainless-steel vacuum chamber.
Cryogenics
operationalHelium liquefaction at 4 K for LTS magnets; HTS magnets relax to 20-40 K, eliminating much of the cryogenic load. Major cost saver.
Diagnostics and Remote Handling
JET remote-handling proven over decades;...Plasma diagnostics (Thomson scattering, interferometry, neutron detection) and in-vessel remote-handling robotics for tritium-contaminated environments.
Ten cross-cutting enabling-technology categories every commercial fusion plant depends on, with primary players, suppliers, maturity, and the single critical gap that gates commercial deployment. HTS magnets unlock compact tokamaks at 20 plus tesla and are the headline enabler. Scientific Q vs engineering Q distinction: NIF reached scientific Q greater than 1 in Dec 2022, but wall-plug Q at NIF is approximately 0.01. Source: FIA Global Fusion Industry Report 2024, CFS HTS magnet test (Sep 2021), ITER Test Blanket Module program, Degrave et al. Nature 602:414-419 (Feb 2022), Sandia Z Machine, LLNL NIF. Accessed 2026-05-15.
Fusion Development Companies
| Company | Country | Founded | Approach | Fuel | Lifetime ($M)▼ | Commercial Target | Investability |
|---|---|---|---|---|---|---|---|
Commonwealth Fusion Systems Devens, MA | United States | 2018 | tokamak | D-T | $2,900 | 2027 (SPARC Q>1), early 2030s (ARC commercial) | private |
TAE Technologies Foothill Ranch, CA | United States | 1998 | FRC | p-B11 (target); D-T intermediate | $1,340 | 2025+ (Copernicus first plasma); FOAK ~2030s | private |
Helion Energy Everett, WA | United States | 2013 | FRC | D-He3 (intermediate D-D) | $1,070 | Polaris electricity demo 2025+ (revised); Microsoft PPA 2028risk | private |
Pacific Fusion Fremont, CA | United States | 2023 | hybrid | D-T | $900 | first ignition ~2030; pilot ~mid-2030s | private |
General Fusion Richmond, BC | Canada | 2002 | MTF | D-T | $325 | 2025-2026 LM26 demonstration (revised); FOAK ~2030srisk | private |
Tokamak Energy Milton Park, Oxfordshire | United Kingdom | 2009 | spherical | D-T | $250 | 2030s commercial prototype; UK STEP 2040 | private |
Marvel Fusion Munich | Germany | 2019 | ICF (indirect) | D-T (early); p-B11 (long-term) | $250 | Research facility late 2020s; commercial 2030s+ | private |
Zap Energy Everett, WA | United States | 2017 | Z-pinch | D-T | $200 | FOAK ~2030s | private |
First Light Fusion Yarnton, Oxfordshire | United Kingdom | 2011 | ICF (indirect) | D-T | $110 | no longer pursuing FOAK plant; licensing technology to partners | private |
Xcimer Energy Denver, CO | United States | 2022 | ICF (direct) | D-T | $109 | pilot 2030s; FOAK ~late 2030s | private |
Type One Energy Madison, WI (HQ); Knoxville, TN (operations) | United States | 2022 | stellarator | D-T | $82.4 | FOAK ~early 2030s | private |
Energy Singularity Shanghai | China | 2021 | tokamak | D-T | $55 | demonstrator 2027+; FOAK 2030s+ | restricted |
Avalanche Energy Tukwila, WA | United States | 2018 | hybrid | D-D (initial); D-He3 / p-B11 long-term | $50 | small-scale power ~late 2020s+ | private |
Realta Fusion Madison, WI | United States | 2022 | stellarator | D-T | $45 | FOAK ~2030s | private |
Proxima Fusion Munich | Germany | 2023 | stellarator | D-T | $35 | demonstrator 2031; FOAK 2030s | private |
Open Star Technologies Wellington | New Zealand | 2021 | tokamak | D-T | $30 | demonstrator 2030s | private |
Renaissance Fusion Grenoble | France | 2019 | stellarator | D-T | $24 | FOAK ~2030s | private |
HB11 Energy Sydney | Australia | 2017 | hybrid | p-B11 | $22 | demonstration 2030s | private |
Focused Energy Darmstadt + Austin TX | Germany / United States | 2021 | ICF (direct) | D-T | $15 | demonstration 2030s | private |
ENN Group fusion Langfang, Hebei | China | 2018 | spherical | p-B11 (target); D-T early | n/a | demonstration 2030s | restricted |
20 of 20 private fusion developers targeting civilian grid power. All companies are privately held; no public-market pure-play fusion exists as of 2026-05-15. Filter by approach or country. Energy Singularity and ENN Group are flagged restricted (China). Lifetime funding triangulates FIA Global Fusion Industry Report 2024, Crunchbase, PitchBook, and company press. Propulsion-focused fusion (Pulsar, Helicity) excluded. Source: FIA, company press, SEC Form D filings.
Capital Concentration by Approach
Top 4 capital concentration
90%
$6.21B in Commonwealth Fusion Systems, TAE Technologies, Helion Energy, Pacific Fusion
Capital concentration by fusion approach across 20 private developers. Tokamak (CFS dominates) plus FRC (TAE + Helion) collectively hold roughly 76 percent of cumulative private fusion capital. Pacific Fusion alone (hybrid pulsed-power Z-pinch ICF) booked the largest Series A in fusion history ($900M Oct 2024). ~90 percent of cumulative private fusion capital sits in 4 companies (CFS, TAE, Helion, Pacific Fusion). Tokamak ~42 percent and FRC ~34 percent dominate by approach. Stellarator + Z-pinch + ICF + MTF + hybrid (excluding Pacific Fusion) collectively account for ~24 percent. Source: derived from nu-fusion-companies.json, FIA Global Fusion Industry Report 2024. Accessed 2026-05-15.
Commercial Targets vs Independent Plausibility
FIA 2024 survey. 70 percent of private fusion companies expect commercial fusion by 2035; majority view is 2040 to 2050. Editor judgment in the chart distinguishes scientific Q greater than 1 (the headline milestone) from FOAK commercial pilot (the grid-power milestone). The two are typically 6 to 10 years apart in fusion roadmaps.
Polaris (electricity demo)
Copernicus (Q greater than 1)
LM26 (demo)
SPARC (Q greater than 1)
ST80-HTS (technology demonstrator) (Q greater than 1)
Colorado State research facility (demo)
HH series follow-on (demonstrator)
Microsoft 50 MW plant (FOAK)
FuZE-Q follow-on (Q greater than 1)
Orbitron compact reactor (demo)
Da Vinci (planned FOAK) (FOAK)
Pulsed-power demonstrator (ignition)
ARC (FOAK)
Infinity One (Bull Run TN) (FOAK)
Pilot plant (pilot)
Commercial pilot (FOAK)
Excimer driver pilot (pilot)
WHAM follow-on (FOAK) (FOAK)
Alpha stellarator (FOAK) (FOAK)
Stellarator with liquid-metal wall (FOAK) (FOAK)
UK STEP partnership (FOAK)
Per-company commercial milestone targets with editor judgment on independent plausibility. Solid filled dot = company-stated year; red-outlined hollow dot = independently-plausible year; gray bar = the gap. Disputed badge flags rows where independent assessment differs materially (Helion Polaris and Microsoft PPA, TAE Da Vinci p-B11, Marvel commercial pilot, CFS ARC, General Fusion LM26). First Light Fusion (Nov 2024 pivot to licensing-only) excluded from the timeline. FIA 2024 industry survey (45 private fusion companies): 70 percent of companies say fusion will provide electricity to the grid by end of 2035; 89 percent by end of 2030s. Editor view: company targets cluster optimistically; the gap to independently-plausible years is the central judgment of this file. Source: company press, FIA Global Fusion Industry Report 2024, UKAEA STEP, Microsoft + Helion PPA (May 2023). Accessed 2026-05-15.
Annual Fusion Investment 2010-2025
Annual fusion investment 2010 to 2025 stacked by source. Private = VC equity into private developers; public = ITER assessed contribution + NIF and LLNL ICF + DOE Office of Fusion Energy Sciences + ARPA-E + DOE Milestone Program; strategic = hyperscaler offtake prepayments (Microsoft + Helion) and corporate strategic equity. 2021 inflection ($3.7B) driven by CFS Series B $1.8B + Helion Series E $500M + General Fusion E. Dec 2022 NIF ignition sustained fundraising into 2022. 2023 cool-down ($1.6B). 2024 recovery ($4.5B): Pacific Fusion Series A $900M + CFS B2 $863M + Helion F $425M. Source: FIA Global Fusion Industry Report 2024, DOE FES appropriations, ITER cash-call schedule, Crunchbase, PitchBook. Accessed 2026-05-15.
Top 20 Private Fusion Equity Rounds
Top 20 disclosed private fusion equity rounds 2018 to 2025. Excludes contingent tranches (Helion Series E $1.7B contingent on Polaris electricity demonstration), pure public grants (DOE Milestone, ARPA-E), and corporate strategic equity reported separately under the investor map and hyperscaler bets tables. Pacific Fusion Series A (Oct 21, 2024) is the largest Series A in fusion history. Source: company press releases, Bloomberg, Reuters, TechCrunch, Crunchbase, PitchBook. Accessed 2026-05-15.
Active Fusion Investors
ENI S.p.A.
corporate$400M
1 portfolio company
Commonwealth Fusion Systems
Sam Altman (individual)
individual$375M
1 portfolio company
Helion Energy
Breakthrough Energy Ventures
venture$350M
5 portfolio companies
Commonwealth Fusion Systems, Pacific Fusion, Type One Energy, Xcimer Energy, Zap Energy
Concentration notes. Broadest portfolio: Lowercarbon Capital with 7 fusion portfolio companies. Largest single deployer (estimate): ENI (~$400M into CFS across rounds + strategic alliance). Unique operator: Sam Altman: only operator-investor staking >$300M personal capital in fusion.
| Investor | Type | Total Fusion ($M) | Portfolio | Companies | Public Thesis |
|---|---|---|---|---|---|
| ENI S.p.A. | corporate | $400 | 1 | Commonwealth Fusion Systems | ENI publicly committed to net-zero by 2050 with fusion as a strategic 2030s-2040s decarbonization play. Largest oil-major equity position in fusion outside Chevron + TAE. |
| Sam Altman (individual) | individual | $375 | 1 | Helion Energy | Altman has stated D-He3 FRC is the fastest path to commercial electricity. Conflict-of-interest considerations: Altman is also CEO of OpenAI, which has electricity demand that could buy Helion output; Microsoft, OpenAI's largest investor, signed first commercial fusion PPA with Helion. |
| Breakthrough Energy Ventures | venture | $350 | 5 | Commonwealth Fusion Systems, Pacific Fusion, Type One Energy, Xcimer Energy, Zap Energy | Climate-stack bets that need to be carbon-negative at gigawatt scale; fusion is the highest-impact electricity bet in BEV's portfolio. |
| Khosla Ventures | venture | $280 | 4 | Commonwealth Fusion Systems, Pacific Fusion, Realta Fusion, Type One Energy | Khosla has stated fusion is the only credible path to base-load clean energy at the scale required for global decarbonization. |
| Tiger Global Management | venture | $250 | 1 | Commonwealth Fusion Systems | Late-stage crossover positioning; Tiger publicly framed fusion as part of its energy-transition deep-tech mandate before pulling back from frontier-tech in 2022-2023. |
| General Catalyst | venture | $250 | 1 | Pacific Fusion | General Catalyst Health Assurance + climate energy thesis; Pacific Fusion is GC's first fusion bet. |
| Lowercarbon Capital | venture | $200 | 7 | Avalanche Energy, Pacific Fusion, Realta Fusion, Renaissance Fusion, Type One Energy, Xcimer Energy, Zap Energy | Diversified fusion bet across approaches; betting that one of seven companies will reach commercial pilot scale. |
| Founders Fund | venture | $110 | 2 | Avalanche Energy, Pacific Fusion | Frontier-physics bets at small check size (Avalanche compact reactor) and large strategic bets (Pacific Fusion pulsed-power). |
| Chevron Technology Ventures | corporate | $90 | 2 | TAE Technologies, Zap Energy | Chevron tech-vc thesis: hydrogen + fusion + carbon-capture portfolio as long-dated optionality alongside core hydrocarbon book. |
| Temasek | sovereign | $90 | 2 | Commonwealth Fusion Systems, General Fusion | Sovereign long-duration clean-energy bet. |
| Bezos Expeditions | family office | $80 | 2 | Commonwealth Fusion Systems, General Fusion | Bezos has publicly framed fusion as a long-term planetary energy bet. General Fusion 2024 funding gap and restructuring is a major position-risk event for this portfolio. |
| Soros Fund Management | family office | $70 | 2 | Commonwealth Fusion Systems, Zap Energy | Climate-aligned long-duration bets. |
| Google Ventures (GV) and Google | corporate | $60 | 2 | TAE Technologies, Commonwealth Fusion Systems | Strategic optionality for data center power demand. |
| Hedosophia | venture | $60 | 1 | Xcimer Energy | Hard-tech moonshot bets across space + energy. |
| Founders + executives via Future Ventures (Steve Jurvetson) | venture | $40 | 1 | Commonwealth Fusion Systems | Frontier deep-tech bet. |
| TDK Ventures | corporate | $25 | 1 | Type One Energy | Magnetics + power electronics supply chain optionality. |
16 active capital allocators with disclosed positions in private fusion-development companies. Total fusion deployment estimates triangulate disclosed round participation, lead status, and pro-rata sharing where individual checks are not disclosed; treat as directional with 50 to 100 percent error bars per investor. Sam Altman is the only operator-investor staking more than $300M of personal capital; conflict-of-interest considerations apply (OpenAI CEO, Microsoft is OpenAI largest backer and Helion PPA counterparty). In-Q-Tel flagged unverified pending public disclosure. Source: FIA Global Fusion Industry Report 2024, BEV portfolio, Khosla Ventures, Lowercarbon Capital, Founders Fund, ENI press, SEC Form D filings. Accessed 2026-05-15.
Hyperscaler and Strategic Corporate Bets
World's first binding fusion PPA
Microsoft + Helion Energy (2023-05-10)
50 MW grid PPA, 2028 target start, world's first commercial fusion PPA. Helion paid early payments; binding penalties if Helion misses delivery.
Skepticism: Helion has missed previous timeline commitments (Polaris electricity demo revised from 2024 to 2025+). Microsoft PPA includes performance ramp and financial penalties on Helion side. Physics community skepticism of 2028 delivery is broad. The PPA is binding electricity-purchase: there is no analogous binding fusion contract.
Binding fusion PPAs
2
Microsoft + Helion (May 10 2023, 50 MW Washington State, 2028 target) and Google + CFS (Jun 30 2025, 200 MW ARC Virginia, early 2030s) are the two binding fusion electricity-purchase agreements. Others are MoU, strategic alliance, equity, or research.
Hyperscaler participation
2 / 5
Microsoft (Helion PPA + equity), Google (CFS PPA + CFS + TAE equity + EPFL research). AWS, Meta, Oracle have none.
Hyperscalers absent
3
Amazon Web Services, Meta Platforms, Oracle have no disclosed fusion position. Negative signal vs aggressive fission deal flow.
Sam Altman conflict triangle. Conflict-of-interest tension: Altman is also CEO of OpenAI, which has electricity demand. Microsoft (OpenAI's largest backer and Helion PPA buyer) signed Microsoft + Helion PPA May 2023. Editorial caveat: this triangle is unusually concentrated and warrants disclosure when discussing Helion timelines.
| Principal | Type | Partner | Nature | Announced | Headline / Skepticism |
|---|---|---|---|---|---|
| Microsoft | hyperscaler | Helion Energy | binding PPA + prepayment | 2023-05-10 | 50 MW grid PPA, 2028 target start, world's first commercial fusion PPA. Helion paid early payments; binding penalties if Helion misses delivery. Skepticism: Helion has missed previous timeline commitments (Polaris electricity demo revised from 2024 to 2025+). Microsoft PPA includes performance ramp and financial penalties on Helion side. Physics community skepticism of 2028 delivery is broad. The PPA is binding electricity-purchase: there is no analogous binding fusion contract. |
| hyperscaler | EPFL Swiss Plasma Center (TCV tokamak) | research collaboration | 2022-02-16 | DeepMind + EPFL Nature paper using reinforcement learning to control plasma shape on TCV tokamak. No equity or PPA. Research collaboration only. Skepticism: Research only, not a commercial bet. Google's commercial fusion position is the corporate row immediately below (TAE equity + CFS equity + Google CFS 200 MW PPA Jun 30 2025). | |
| Google (corporate) | hyperscaler | TAE Technologies + Commonwealth Fusion Systems | equity (TAE + CFS) + binding... | 2022-07-19 (TAE Series G); 2025-06-30 (CFS PPA + expanded CFS equity); 2025-08-28 (CFS B2 participation) | Google co-led TAE Series G Jul 2022 with Chevron and Sumitomo. On Jun 30 2025, Google signed a binding 200 MW PPA with CFS for the ARC plant in Chesterfield County, Virginia (early 2030s target; largest direct corporate fusion offtake to date) plus increased CFS equity. Google has option to offtake from additional ARC plants. Google also participated in CFS Series B2 Aug 28 2025. Skepticism: Google + CFS Jun 30 2025 PPA is the second binding fusion electricity-purchase agreement (after Microsoft + Helion May 10 2023). Early-2030s ARC timing carries significant slip risk per editor judgment putting ARC at 2036+. |
| Amazon Web Services | hyperscaler | none | none disclosed | n/a | AWS has no disclosed fusion position as of data cutoff. All Amazon nuclear bets (Talen, X-energy, Energy Northwest, Dominion) are fission. Amazon strategy appears to be early-commercial advanced fission rather than fusion optionality. Skepticism: Negative signal: AWS has been most aggressive hyperscaler on nuclear but explicitly chose fission. Worth tracking whether AWS adds fusion position in 2026-2027. |
| Meta Platforms | hyperscaler | none | none disclosed | n/a | Meta has no disclosed fusion position. Constellation Clinton PPA Jun 2025 is fission. No public fusion strategy. Skepticism: Negative signal. Meta clean-energy strategy is renewables + fission existing fleet. |
| Oracle | hyperscaler | none | none disclosed | n/a | Oracle SMR commentary (Sep 2024 + Sep 2025 earnings calls) references fission SMRs, not fusion. No fusion position disclosed. |
| Jeff Bezos (personal, via Bezos Expeditions) | individual / family ... | General Fusion + Commonwealth Fusion Systems | equity (multiple rounds) | 2011 onward | Bezos invested in General Fusion via Bezos Expeditions starting ~2011 (multiple rounds); participated in CFS Series B Dec 2021. Estimated cumulative $40-80M. Skepticism: Bezos position separate from Amazon Web Services nuclear strategy. General Fusion is currently in financial restructuring (Sep 2024 layoffs). |
| Bill Gates (personal, via Breakthrough Energy Ventures + direct) | individual / family ... | Commonwealth Fusion Systems | equity (BEV lead + direct) | 2018 onward | Gates was direct participant in CFS Series B Dec 2021 alongside BEV's lead position. Estimated direct check $25-50M, plus BEV's ~$50M+ across rounds. Skepticism: Gates has named CFS as the BEV fusion bet. Gates also funds TerraPower (fission SMR), so fusion is not exclusive bet. |
| Sam Altman (personal) | individual (operator... | Helion Energy | Chairman + lead investor | 2014 onward; led Series E Nov 2021 + Series F mid-2024 | Altman is Helion Chairman; led Series E with reported $375M personal check; led Series F. Total personal investment exceeds $400M. Unique in fusion: no other operator has staked this fraction of personal balance sheet. Skepticism: Conflict-of-interest tension: Altman is also CEO of OpenAI, which has electricity demand. Microsoft (OpenAI's largest backer and Helion PPA buyer) signed Microsoft + Helion PPA May 2023. Editorial caveat: this triangle is unusually concentrated and warrants disclosure when discussing Helion timelines. |
| ENI | corporate strategic ... | Commonwealth Fusion Systems | strategic alliance + equity | 2018 onward; reinforced Oct 2024 | ENI strategic alliance with CFS includes equity (Series B + B2), industrial scaling partnership, and exploratory off-take for European pilot. Estimated cumulative position $300-500M including program-level commitments. Largest oil-major fusion bet. Skepticism: ENI net-zero by 2050 strategy puts fusion in the 2030s-2040s decarbonization path. Strategic alliance is not a binding PPA but is the deepest industrial relationship in fusion. |
| TVA (Tennessee Valley Authority) | utility (US federal) | Type One Energy | MoU + site host designation | 2023-05 | TVA + Type One Energy MoU May 2023 designating Bull Run TN as host site for Infinity One fusion plant. TVA is a US federal utility; this is the most concrete utility-developer partnership in fusion. Not a binding PPA. Skepticism: MoU only, FOAK target early 2030s. TVA Bull Run is a retired coal plant site offering grid connection and water access. Site-host designation is unusual depth for an MoU. |
| Chevron Technology Ventures | corporate strategic ... | TAE Technologies + Zap Energy | equity | 2014 (TAE) onward; Zap Series D Jul 2024 | Chevron Technology Ventures co-led TAE Series G Jul 2022; participated in Zap Energy Series D Jul 2024. Total exposure estimated $50-100M. Skepticism: Strategic R&D positioning, not industrial alliance scale of ENI + CFS. |
Hyperscaler and strategic corporate positions in fusion specifically. Only Microsoft + Helion (May 10, 2023) is a binding electricity-purchase agreement; all other relationships are MoU, strategic alliance, equity, or research. AWS, Meta, Oracle all have no disclosed fusion position as of 2026-05-15 (negative signal vs aggressive advanced-fission deal flow). Sam Altman + OpenAI + Microsoft + Helion concentration is unique in private fusion and warrants disclosure when discussing Helion timelines. Source: Microsoft + Helion PPA announcement, DeepMind + EPFL Nature paper (Feb 2022), TVA + Type One MoU (May 2023), ENI + CFS strategic alliance (Oct 2024), Bezos Expeditions, Bill Gates blog, SEC Form D filings.
Scientific Milestones Timeline
Critical Q distinction. NIF Dec 5 2022 ignition is Q_laser (3.15 MJ fusion from 2.05 MJ laser), not Q_wall-plug. Q_wall-plug at NIF is ~0.01 because lasers run ~1 percent wall-plug efficient (300 MJ grid for one 2.05 MJ pulse). JET Q_thermal 0.67 (1997) is the highest sustained MCF Q. ITER targets Q_thermal = 10. Commercial fusion requires Q_wall-plug greater than ~3.
Major fusion scientific and engineering milestones 1991 to 2025. Mix of public-program (JET, NIF, EAST, KSTAR, W7-X, JT-60SA) and private-developer milestones (Helion Trenta, CFS HTS magnet, ST40, Energy Singularity HH70). Inflection markers: JET 16.1 MW DT-1 (1997), JET DT-2 plus NIF ignition (2022), EAST 403s plus W7-X 30-minute plasma (2023). Sources: LLNL, EUROfusion, IPP CAS, KFE, IPP Greifswald, QST, UKAEA, MIT News, Tokamak Energy press.
Public Mega-Programs
ITER cost range $25B to $65B. The lower figure ($25B) reflects member-assessed contributions in nominal dollars; the upper figure ($65B) is the US DOE 2018 life-cycle replacement value. Both are defensible. Press tends to cite the upper figure. ITER is a research facility, not a power plant: it produces zero grid electricity.
Intergovernmental2 programs
| Program | Host | Approach | Constr. start | Revised schedule | Funding ($B) | Status |
|---|---|---|---|---|---|---|
ITER not a power plant | Cadarache, France (international consortium of 35 countries) | MCF tokamak (D-T capable) | 2010 | Nov 2024 ITER Council revision: Start of Research Operation (H + D-D plasmas) 2034; full magnetic energy 2036; DT operation 2039 | 25 to 65 | Construction ongoing; cryostat complete; vacuum vessel sector assembly in progress; tokamak first sector complete Apr 2023 |
JT-60SA not a power plant | QST Naka, Japan (EU-Japan Broader Approach) | MCF superconducting tokamak (ITER-class) | 2007 | Operating; commissioning ongoing 2024-2025 | 1 | World's largest operational superconducting tokamak (until ITER) |
EU and UK2 programs
| Program | Host | Approach | Constr. start | Revised schedule | Funding ($B) | Status |
|---|---|---|---|---|---|---|
DEMO (EUROfusion) commercial-prototype target | EU EUROfusion consortium | MCF tokamak (D-T capable, commercial demonstrator) | design phase | Roadmap reviewed periodically; 2050+ commercial demonstration target | 0.5 | Pre-conceptual design; awaits ITER operational results |
UK STEP (Spherical Tokamak for Energy Production) commercial-prototype target | UKAEA, West Burton, UK | MCF spherical tokamak (D-T capable, commercial prototype) | 2027 | On schedule as of Sep 2024 announcement; £220M Phase 1 budget committed | 0.3 | Concept design phase; Tokamak Energy + EDF + Atkins consortium |
National8 programs
| Program | Host | Approach | Constr. start | Revised schedule | Funding ($B) | Status |
|---|---|---|---|---|---|---|
CFETR (China Fusion Engineering Test Reactor) commercial-prototype target | China Academy of Sciences IPP | MCF tokamak (D-T capable, commercial demonstrator) | design phase | Roadmap reviewed in 14th + 15th Five-Year Plans | 6 | Engineering design phase; CN partner of ITER and post-ITER bridge to commercial |
JET (Joint European Torus) not a power plant | Culham, UK (decommissioned Dec 2023) | MCF tokamak (D-T capable) | 1978 | Decommissioned Dec 2023 after JET DT-2 campaign; final DT operation completed | 4.5 | Decommissioning underway; data and methodology informing ITER + UK STEP |
NIF (National Ignition Facility) not a power plant | Lawrence Livermore National Laboratory, US | ICF indirect drive (X-ray drive, 192 laser beams) | 1997 | Mission complete on ignition basis; current research phase optimizes yield + reproducibility | 3.5 | Operating since 2009; ignition achieved Dec 2022; peak yield 5.2 MJ Feb 2024 |
Wendelstein 7-X not a power plant | IPP Greifswald, Germany | MCF stellarator (quasi-isodynamic) | 1996 | Ongoing operations; not D-T capable (deuterium-only) | 1.5 | Operational; longest hot plasma in any fusion machine |
EAST not a power plant | Hefei IPP, Chinese Academy of Sciences | MCF superconducting tokamak | 2003 | Ongoing operations; 403s plasma Apr 12 2023 record; 1,066s H-mode plasma Jan 20 2025 (subsequently surpassed by WEST 1,336s Feb 18 2025) | 0.4 | Operational; produces sustained-plasma records |
KSTAR not a power plant | KFE Daejeon, South Korea | MCF superconducting tokamak | 1995 | Ongoing operations; 300s plasma target | 0.35 | Operational; tungsten divertor upgrade enabled higher-temperature plasmas |
HL-3 (formerly HL-2M) not a power plant | Southwestern Institute of Physics (SWIP), Chengdu, China | MCF tokamak | 2009 | Operating | 0.2 | Operational; supports China CFETR design |
DOE Milestone-Based Fusion Development Program commercial-prototype target | US Department of Energy + 8 private companies | Mixed (technology-agnostic) | 2023 | On schedule | 0.046 | 8 companies receiving milestone-based grants; ARPA-E + DOE oversight |
Major public fusion mega-programs, national laboratory facilities, and intergovernmental projects. Excludes private-developer machines. Funding figures are cumulative multi-year commitments. ITER cost range ($25B member-assessed to $65B DOE life-cycle) reflects accounting basis. JET decommissioned Dec 2023 after the DT-2 record campaign. UK STEP and EU DEMO and China CFETR are the public-sector commercial-prototype targets (all pre-construction). DOE Milestone Program ($46M across 8 companies) is modest in dollar terms but structurally important. Sources: ITER Organization, US DOE, UKAEA, EUROfusion, QST, IPP CAS, KFE, SWIP.
ITER Status Tracker
ITER is NOT a power plant. It produces zero grid electricity. ITER is a research facility designed to demonstrate Q_thermal = 10 sustained for 400 seconds at 500 MW thermal output. Q_thermal is fusion power vs plasma-heating power, not vs grid input. Post-ITER, a commercial demonstrator (EUROfusion DEMO, planned 2050s+) would add a thermal-to-electric system, tritium breeding blanket, and steady-state operation.
Cryostat
complete
1,250 tonnes, Apr 2020
First plasma
n/a
14-year slip vs 2007 baseline
DT operation
2039
deuterium-tritium
Cost range
$25-65B
member-assessed vs DOE life-cycle
First-plasma forecast slip, 2007 to Nov 2024
| Review | First plasma forecast | DT operation forecast |
|---|---|---|
| 2007 | 2020 | 2035 |
| 2010 | 2019 | 2026 |
| 2016 | 2025 | 2035 |
| 2022 | 2025 | 2035 |
| 2024-11 | 2034 (Start of Research Operation, H + D-D) | 2039 |
| Milestone | Original | Nov 2024 revised | Status | Slip (yr) | Notes |
|---|---|---|---|---|---|
| Construction site preparation | 2007 | complete | complete | . | Cadarache site cleared; first concrete poured 2010 |
| Tokamak Complex building | 2017 | complete 2020 | complete | . | Three buildings, tokamak pit, 20 levels |
| Cryostat installation | 2019 | complete 2020 | complete | . | Cryostat base lowered Apr 2020; one of the largest single components ever installed (1,250 tonnes) |
| Vacuum vessel sector assembly | 2020-2022 | in progress, completion 2026-2028 | in progress | . | 9 vacuum vessel sectors; manufacturing defects in some sectors require repair; major schedule risk source |
| Tokamak first sector installation | 2021 | complete Apr 2023 | complete | . | First of nine sectors installed Apr 2023; remaining sectors continue assembly |
| Toroidal field coils delivery | 2020-2021 | complete 2020-2024 | complete | . | 18 toroidal field coils manufactured by EU (10) + Japan (8); largest superconducting magnets ever built |
| Cooling water system | 2023 | in progress, completion 2028+ | in progress | . | Component-specific timelines |
| Start of Research Operation (hydrogen + D-D plasmas) | 2020 (first plasma) | 2034 | delayed | 14 | Original 2020 first-plasma was set in 2007 baseline; slipped to 2025 in 2016 review; Nov 2024 ITER Council revision redefined this gate as Start of Research Operation (hydrogen and D-D) in 2034. ~14-year slip vs original. |
| Full magnetic energy | 2033 (per 2016 reference plan) | 2036 | delayed | 3 | Nov 2024 ITER Council baseline: full magnetic energy 2036, a 3-year delay vs the 2016 reference plan. |
| Deuterium-tritium operation | 2035 | 2039 | delayed | 4 | Nov 2024 ITER Council baseline pushes DT to 2039, a 4-year delay vs the previous reference. |
| Full deuterium-tritium baseline performance (Q_thermal = 10) | ~2035 | post-2039 | delayed | 5 | Q_thermal = 10 means fusion power output is 10x plasma-heating input. NOT Q_wall-plug; still does not feed grid. Full Q=10 phase follows the 2039 DT start. |
ITER (Cadarache, France) is the world's largest fusion construction project, host-funded by 35 countries. The Nov 2024 baseline revision is the most consequential schedule slip in fusion history: first plasma moved from 2020 (2007 baseline) to 2034-2035, a 14-year slip. DT operation pushed from 2035 to 2039+. Cost range $25B (member-assessed nominal) to $65B (DOE life-cycle replacement). Sources: ITER Organization newsline, US DOE 2018 ITER review, GAO ITER cost reports.
NIF Yield History
Q_laser is not Q_wall-plug. Q_laser is fusion-energy-out / laser-energy-in. Q_laser > 1 was achieved Dec 5 2022 with 3.15 MJ / 2.05 MJ = 1.54. Q_wall-plug is fusion-energy-out / grid-electricity-in. NIF requires ~300 MJ of grid electricity to deliver a 2.05 MJ laser pulse, because excimer/Nd:glass laser wall-plug efficiency is ~1 percent. Therefore Q_wall-plug at NIF Dec 2022 was 3.15 / 300 ~ 0.01. Commercial fusion requires Q_wall-plug > ~3 after accounting for thermal-to-electric conversion (~40 percent efficient) and recirculating power. NIF is 300x below commercial threshold on wall-plug basis. NIF shot rate is ~1 shot per day with extensive between-shot target swap and laser-system reset. Commercial ICF requires ~10 Hz (~10 shots per second) sustained, never demonstrated.
NIF (National Ignition Facility, LLNL) fusion yield per major disclosed shot 2009 to 2025. Y-axis is fusion energy delivered to the target chamber (MJ). The dashed line marks the Q_laser = 1 ignition threshold at 2.05 MJ (laser energy delivered). First ignition Dec 5 2022 at 3.15 MJ (Q_laser 1.54), reproduced Jul 2023 (3.88 MJ), peak Feb 2024 (5.2 MJ). Q_wall-plug remains ~0.01 because lasers run ~1 percent wall-plug efficient. NIF shot rate is ~1 per day; commercial ICF requires ~10 Hz sustained. Sources: LLNL press, Physical Review Letters Feb 2024 ignition paper, DOE ICF program review.
Regulatory Framework by Country
US NRC Apr 14 2023 (key precedent). The NRC voted to regulate fusion energy machines under 10 CFR Part 30 (byproduct material / particle accelerator), NOT Part 50/52 (utilization facility, the fission framework). The result: light-touch regime, no NRC reactor license needed, state-level oversight handles byproduct material and worker safety. This removes 5 to 10 years of fission-style pre-construction licensing from the commercial timeline and is the single biggest US regulatory enabler for private commercial fusion. UK Energy Act 2023 and Japan METI 2023 strategy adopted parallel light-touch positions, forming a US-UK-Japan trilateral consensus.
| Country | Regulator | Current framework | Key decision | Status | Implications |
|---|---|---|---|---|---|
United States key precedent | Nuclear Regulatory Commission (NRC) | 10 CFR Part 30 (byproduct material / particle accelerator) | Apr 14 2023 NRC vote: fusion energy machines regulated under 10 CFR Part 30, NOT Part 50/52 (utilization facility) | in force | Light-touch framework; no NRC reactor license needed; state-level licensing handles radioactive material handling and worker safety. Removes 5-10 years of pre-construction licensing from the typical fission timeline. The single biggest US regulatory enabler for private commercial fusion. |
United Kingdom | Office for Nuclear Regulation (ONR) + Environment Agency (EA) | UK Energy Act 2023 | UK Energy Act 2023: fusion explicitly excluded from nuclear-installation regulatory regime; ONR + EA share fusion oversight under a less-onerous framework | in force | UK STEP and Tokamak Energy benefit from streamlined consent process; modeled on US approach but with environmental authority shared with EA |
Japan | METI (Ministry of Economy, Trade and Industry) + NRA (Nuclear Regulation Authority) | METI 2023 Fusion Energy Innovation Strategy | Apr 2023 METI strategy: light-touch framework modeled on US direction; NRA retains oversight of radioactive materials | in force | JT-60SA operating under fusion-specific regulatory arrangement; Japanese private fusion (Helical Fusion, Kyoto Fusioneering) benefit |
European Union | Member-state-specific (no harmonized framework) | No harmonized EU fusion framework | EU Strategic Energy Technology Plan acknowledges fusion separately from fission; member states retain individual licensing authority | in progress | Germany, France, Italy each have national approaches; ITER (Cadarache, France) operates under French nuclear safety authority ASN with bespoke arrangement. Marvel Fusion (Germany) and Renaissance Fusion (France) navigate national rules. |
Canada | Canadian Nuclear Safety Commission (CNSC) | CNSC fusion-specific Class II Nuclear Facility framework | CNSC determined LM26 (General Fusion) qualifies as Class II Nuclear Facility (lower risk than Class I fission) | in force | General Fusion's LM26 demonstration at Richmond BC operates under Class II permit; lighter than fission Class I but more burdensome than US Part 30 |
South Korea | Nuclear Safety and Security Commission (NSSC) + KFE | State-led research framework | K-DEMO roadmap; KSTAR operates under research permission | state-led | State-led; private fusion development limited |
China | National Nuclear Safety Administration (NNSA) / China Atomic Energy Authority (CAEA) | State-led; no commercial private fusion regulatory framework | n/a; fusion treated under state nuclear research framework | state-led | Energy Singularity, ENN Group fusion operate under state research permissions; commercial framework will likely follow state-led model |
India | Atomic Energy Regulatory Board (AERB) + Department of Atomic Energy (DAE) | State-led; no commercial private fusion framework | n/a; India is ITER member, fusion treated as research | state-led | No private commercial fusion development |
Australia | Australian Radiation Protection and Nuclear Safety Agency (ARPANSA) | Framework TBD | Australian Commonwealth nuclear prohibition (1999) restricts fission; fusion legal status ambiguous | ambiguous | HB11 Energy operating under research permissions; commercial path requires legal clarification |
Russia | Rostechnadzor | State-led; sanctions limit Western participation | n/a | state-led | Russia retains ITER membership; sanctioned Western commercial engagement limited |
Fusion regulatory frameworks by country and region (civilian power scope). The US NRC decision of Apr 14 2023 (10 CFR Part 30, NOT Part 50/52) is the foundational US private-investment enabler, cited in every major US private fusion company investor deck. UK Energy Act 2023 and Japan METI 2023 strategy form a parallel light-touch trilateral consensus. EU has no harmonized framework (drag vs US/UK). Canada CNSC operates a Class II Nuclear Facility framework (intermediate). Australia 1999 nuclear prohibition leaves fusion legal status ambiguous. China, Korea, India, Russia are state-led with no private commercial framework. Sources: NRC SECY-23-0001, UK ONR, METI Japan, EU SET Plan, CNSC, ARPANSA, FIA regulatory tracker.
Fusion FOAK Timeline
Disputed timelines
6 of 16
FIA survey by 2035
70%
FIA survey by 2040-2050
89%
Per-company Gantt-style fusion commercialization timeline: criticality / Q greater than 1 (green), pilot plant (amber), FOAK grid plant (violet). Solid markers are company-stated years; dashed-outline markers show editor independently-plausible years for rows where the stated target is judged optimistic by 4 plus years. FIA 2024 industry survey: ~25 percent of companies expect commercial fusion by 2035, ~50 percent by 2040-2050. Sources: company technology pages (CFS, Helion, TAE, Tokamak Energy, Pacific Fusion, Type One, etc.), FIA Global Fusion Industry Report 2024.
Fusion PPA Pipeline
2 binding electricity-purchase contract in fusion. Microsoft + Helion (May 10 2023, 50 MW Washington State, 2028 target) and Google + CFS (Jun 30 2025, 200 MW ARC Virginia plant, early 2030s) are the two binding electricity-purchase agreements, now both present as rows in this file. ENI + CFS is binding strategic alliance with equity + industrial commitments but no binding electricity offtake. TVA + Type One is binding MoU on site-host but no offtake. Nucor + Helion is collaboration + $35M equity but non-binding offtake. All others removed or downgraded. Press coverage frequently miscites MoU + LOI as PPA. Verify any new claims before treating as binding.
Commercial agreements signed for fusion electricity (fusion-only scope; fission hyperscaler deals are tracked separately). Microsoft + Helion May 10 2023 is the only binding electricity-purchase contract in fusion: 50 MW Washington State, target 2028, includes early payment and financial penalties on Helion if delivery slips. ENI + CFS Oct 2024 is a binding strategic alliance with equity and industrial scaling commitments, no binding electricity offtake. TVA + Type One May 2023 is a non-binding MoU with site-host designation at Bull Run TN. Press coverage frequently miscites MoUs and LOIs as PPAs; only Microsoft + Helion qualifies. Sources: Helion press, Type One press, ENI press, Nucor LOI, FIA pipeline.
Fusion Bull vs Bear Synthesis
Editor synthesis. The fusion bull case is real but fragile. Scientific feasibility is largely demonstrated. Capital is flowing. Regulatory framework is favorable in US/UK/Japan. The bear case concentrates on engineering tractability of Q_wall-plug > 1, closed tritium cycle, ICF shot rate, and the gap between research-Q and commercial-Q. Editor judgment: commercial fusion electricity to grid in any form by 2030 is highly unlikely; by 2035 is possible but not likely; by 2040-2045 is more credible. The single highest-leverage development would be a demonstration of Q_wall-plug > 1 in any private machine.
Scientific feasibility
Bull case
NIF achieved ignition (Q_laser > 1) Dec 5 2022 with 3.15 MJ from 2.05 MJ laser. JET DT-2 set 59 MJ over 5s record Feb 2022. EAST 403s steady plasma Apr 2023. KSTAR 100 M-degree C / 30s. ST40 100 M-degree C in compact spherical. Helion Trenta 100 M-degree C FRC. All key physics thresholds have been crossed in some configuration.
Bear case
NIF Q_wall-plug ~0.01 (laser wall-plug efficiency ~1 percent). JET Q_thermal 0.67 maximum. ITER Start of Research Operation pushed to 2034 (and DT to 2039) in Nov 2024 ITER Council baseline revision, vs original 2020 first plasma. No machine has demonstrated Q_thermal > 1 sustained. Aneutronic fuels (p-B11, D-He3) untested for net energy.
Inflection to watch
First sustained Q_thermal > 1 in any machine (SPARC target 2027, ITER DT 2039, Tokamak Energy ST80-HTS 2027) is the next major threshold.
HTS magnet readiness
Bull case
CFS SPARC TFMC achieved 20 T peak field Sep 2021 at fusion-relevant scale. Tokamak Energy ST40 records. Energy Singularity HH70 first plasma Jun 2024 (second private HTS tokamak). Type One Energy + Proxima Fusion + Realta Fusion scaling. HTS path unlocked compact-scale fusion: SPARC at 1.85 m major radius vs ITER 6.2 m.
Bear case
HTS tape supply chain is constrained. No commercial-scale (40+ T sustained) fusion magnet built. ReBCO tape cost still $50-100/kA-m vs target $25/kA-m. ITER uses Nb3Sn low-temperature superconductors at 11.8 T peak; HTS extrapolation untested at scale.
Inflection to watch
SPARC full magnet integration (2026-2027) and first plasma; ST80-HTS commissioning (Tokamak Energy ~2027).
Tritium supply
Bull case
Lithium blanket breeding (Li-6 + n -> T + He) is being engineered into ARC + DEMO designs. ITER will trial breeding blanket modules. Helion D-He3 path eliminates tritium requirement entirely (uses D-D and harvests He-3 from D-D reactions). HB11 and TAE pursue p-B11 aneutronic path.
Bear case
Global commercial tritium supply ~3 kg/yr (mostly from CANDU heavy water plants). A 1 GWe D-T fusion plant needs ~150 kg/yr tritium. Closed-cycle tritium breeding at scale never demonstrated. Tritium permeates metals; recovery efficiency from blanket is unproven. Aneutronic fuels need temperatures 5-10x higher than D-T.
Inflection to watch
ITER breeding blanket trial results (~2040s). First credible closed-cycle tritium demo at scale would change the bear case.
Capital availability
Bull case
$11-13B cumulative private equity through 2025 (Crunchbase / individual round tally). Hyperscaler strategic interest (Microsoft + Helion PPA, Google + CFS 200 MW PPA Jun 30 2025, Google + TAE equity). Oil major positioning (ENI strategic alliance with CFS, Chevron in TAE + Zap). 2024: Pacific Fusion $900M + Helion F $425M. 2025: CFS Series B2 $863M Aug 28 with Nvidia, Google, Bill Gates, BEV, Khosla, ENI as co-investors.
Bear case
Each company needs $1-5B+ more capital to reach FOAK. ARDP-equivalent program for fusion does not exist ($46M DOE Milestone Program is modest by comparison to $3.5B ARDP for fission SMRs). Capital cycles can dry up (2023 cool-down to $1.4B private). General Fusion Sep 2024 restructuring shows even well-known names can fail to raise.
Inflection to watch
(a) Whether 2025-2026 cumulative private equity hits $20B+; (b) Whether DOE Milestone Program scales to $1B+ Phase 2.
Commercial timeline
Bull case
FIA 2024 survey of 45 private companies: 70 percent expect commercial fusion electricity to grid by end of 2035; 89 percent by end of 2030s. Microsoft + Helion PPA targets 2028. Google + CFS 200 MW PPA Jun 30 2025 targets early 2030s. CFS SPARC Q>1 target 2027. UK STEP 2040 commercial prototype with £220M Phase 1 budget.
Bear case
Independent expert view: commercial fusion at scale 2040-2050. ITER slip pattern (14 years on first plasma, redefined as Start of Research Operation 2034 in Nov 2024 baseline). Helion timeline revisions (Polaris 2024 -> 2025+). p-B11 timelines (TAE Da Vinci 2030) widely judged optimistic. Independent editor judgment puts CFS ARC 2036+, Helion FOAK 2032+, most others 2035-2042. Company-stated targets in FIA survey are self-reported by developers with capital-raise incentive to claim near-term commercialization.
Inflection to watch
Whether any private developer demonstrates Q_thermal > 1 by 2028. Whether Helion delivers Polaris electricity demo by 2027.
Regulatory framework
Bull case
NRC Apr 14 2023 decision: fusion regulated under 10 CFR Part 30, NOT Part 50/52. UK Energy Act 2023 parallel light-touch. Japan METI 2023 strategy aligned. US-UK-Japan trilateral consensus removes 5-10 years of pre-construction licensing from typical fission timeline.
Bear case
Regulatory framework untested at grid-connected commercial scale. State-level oversight (US Agreement States) varies. EU has no harmonized framework. China + Korea + Russia state-led. Australian framework ambiguous. First commercial plant siting + permitting process will set precedent and could surface new issues.
Inflection to watch
First commercial fusion plant site permit (likely Microsoft + Helion Washington State 2026-2027). First state-level Agreement State licensing of fusion handling.
Aneutronic fuel viability (p-B11, D-He3)
Bull case
TAE (p-B11), Helion (D-He3), HB11 (p-B11) actively pursuing. Aneutronic eliminates tritium supply problem, neutron-damage problem, and most radioactive waste. Helion claims D-He3 can directly convert charged particles to electricity (no thermal cycle).
Bear case
p-B11 ignition temperature ~1,500 M-degree C (10x D-T). D-He3 ignition ~500 M-degree C (3-4x D-T). Never demonstrated net energy in any aneutronic fuel. He-3 is scarce on Earth (~20 kg/yr global production). Helion path harvests He-3 from D-D side reactions, but yields are challenging.
Inflection to watch
Any private aneutronic developer reaching Q_thermal > 0.1 in p-B11 or D-He3 would substantially update the bull case.
ICF vs MCF horse race
Bull case
NIF demonstrates ICF physics works. Marvel Fusion + Xcimer Energy + Pacific Fusion building commercial ICF drivers. ICF has cleaner physics (no plasma confinement instabilities) and recent yield momentum (Q_laser 2.5 at Feb 2024 NIF shot).
Bear case
ICF requires 10 Hz drivers (NIF runs ~1 shot per day). Excimer-laser commercial driver never built. Hohlraum target fabrication at scale unsolved. MCF (tokamak + stellarator) has 60+ years of physics maturity vs ICF's 30. Most private capital still flows to MCF (CFS, Helion, Tokamak Energy, Type One vs ICF Marvel + Xcimer + Pacific).
Inflection to watch
Whether any ICF developer demonstrates 1 Hz shot rate with sustained yield. Pacific Fusion 10-year roadmap milestones.
Hyperscaler pull-through
Bull case
Two binding fusion PPAs: Microsoft + Helion May 10 2023 (50 MW Washington State, 2028) and Google + CFS Jun 30 2025 (200 MW ARC Virginia, early 2030s). Google also has TAE equity + EPFL DeepMind research collaboration + expanded CFS equity in Series B2 Aug 2025. Microsoft has Helion PPA + earlier Helion equity. Hyperscaler clean-energy demand is structural and growing (data center load 4-9 percent of US grid demand by 2030).
Bear case
Only 2 binding PPAs in fusion. AWS, Meta, Oracle have zero fusion bets (all chose fission). Non-binding MoUs (TVA Type One, ENI CFS) dominate. Helion timeline-slippage risk threatens the Microsoft PPA's 2028 delivery; CFS ARC PPA timing (early 2030s) carries significant slip risk per editor judgment putting ARC at 2036+.
Inflection to watch
Whether AWS or Meta sign first fusion PPA. Whether Helion delivers Microsoft 2028 obligation or pays penalties. Whether CFS ARC delivers on Google PPA early 2030s target.
Geopolitical positioning
Bull case
US-UK-Japan light-touch regulatory consensus. US private fusion equity dominates (CFS, Helion, TAE, Pacific Fusion). UK STEP government commitment. EU continued ITER investment + DEMO planning. China state-led parallel program but investability restricted.
Bear case
China CFETR roadmap runs ahead of EU DEMO. China state capital subsidization can undercut Western private timelines. Sanctioned Russia retains ITER membership. Australia legal ambiguity. US private sector momentum could be vulnerable to capital cycle.
Inflection to watch
China CFETR construction start (~2030). US continued private capital flow.
critical unknowns4 open questions
When (if ever) does any machine demonstrate sustained Q_wall-plug > 1?
Why it matters: Q_laser > 1 (NIF) and Q_thermal targets at JET/ITER are both compatible with Q_wall-plug ~0.01. Commercial fusion requires Q_wall-plug > ~3. No public timeline targets Q_wall-plug > 1 before mid-2030s, and no roadmap demonstrates the path is engineering-tractable.
Earliest resolution: 2028 if Helion delivers Polaris electricity demo; 2032+ if Helion slips and CFS SPARC follow-on Q_thermal > 1 doesn't have favorable wall-plug economics; 2040+ if ITER-class is the path.
Can closed-cycle tritium breeding be demonstrated at scale?
Why it matters: D-T fusion needs ~150 kg/yr tritium per GWe; global commercial supply is ~3 kg/yr. No breeding blanket has run at scale. Failure here forces aneutronic-only path or makes commercial D-T uneconomic.
Earliest resolution: ITER breeding blanket modules trial 2040s+. Private developer breeding tests (CFS ARC, UK STEP) earlier but at smaller scale.
Can ICF achieve 10 Hz commercial shot rate?
Why it matters: NIF runs 1 shot per day. Commercial ICF needs ~10 Hz (10 shots per second). Excimer-laser commercial driver, hohlraum target fabrication, and target-injection rate are all untested at scale.
Earliest resolution: Pacific Fusion + Xcimer Energy 10-year roadmaps target ~2033-2035 pilot demonstrations. Failure could shift the bear case decisively against ICF.
Does the US light-touch regulatory framework survive first commercial siting?
Why it matters: NRC Apr 2023 decision is the foundation of US private investment thesis. First commercial plant siting will test state-level Agreement State permitting, water rights, environmental review. New issues could emerge that the framework was not designed for.
Earliest resolution: First commercial site permit (Microsoft + Helion 2026-2027) is the first stress test. Outcome shapes 2027-2032 regulatory trajectory.
Fusion-specific bull-vs-bear synthesis across 10 dimensions (scientific feasibility, HTS magnet readiness, tritium supply, capital availability, commercial timeline, regulatory, aneutronic fuels, ICF vs MCF, hyperscaler pull-through, geopolitical positioning) plus 4 critical unknowns. Editor judgment: commercial fusion electricity by 2030 is highly unlikely; by 2035 is possible but not likely; by 2040-2045 is more credible. The single highest-leverage development would be a demonstration of Q_wall-plug greater than 1 in any private machine. Sources cross-reference nu-fusion-scientific-milestones.json, nu-fusion-iter-status.json, nu-fusion-nif-yield-history.json, nu-fusion-foak-timeline.json, nu-fusion-regulatory.json, nu-fusion-ppa-pipeline.json, plus FIA Global Fusion Industry Report 2024.