Ten major developments have reshaped the fusion landscape since our February 2026 baseline.
The 10 companies defining the commercial fusion race, from the most-funded to the most unconventional.
Seven distinct approaches to commercial fusion, compared across key dimensions.
| Approach | Mechanism | Leading Companies | Pros | Cons | Status | Timeline |
|---|---|---|---|---|---|---|
| Tokamak MCF |
Toroidal plasma confined by external magnets + plasma current | CFS, ITER, China EAST | Most studied (70+ yrs), HTS magnets shrink size, SPARC 2027 | Complex, expensive, disruption risk, needs current drive | Construction (SPARC) | 2027–2035 |
| Stellarator MCF |
Twisted toroidal field — no plasma current needed | Proxima Fusion, Type One | Inherently steady-state, no disruptions, stable | Extremely complex coil geometry, harder to build, less studied | R&D → Alpha MOU | 2030s |
| Field-Reversed Config. FRC |
Self-contained plasma ring, pulsed or beam-injected | Helion, TAE | Compact, direct electricity (Helion), aneutronic potential (TAE) | Stability challenges, less physics validation, shorter confinement | Prototype ops | 2026–2028 |
| Inertial Confinement ICF |
Lasers compress fuel pellet to extreme density + temperature | Inertia Enterprises, NIF | Only approach with proven ignition (NIF 2022), high energy density | Extremely low repetition rate, massive laser infrastructure, efficiency | Early commercial R&D | TBD (long) |
| Magnetized Target MTF |
Liquid metal pistons compress magnetized plasma | General Fusion | No extreme conditions, liquid metal = blanket + coolant, simpler | Piston synchronization, plasma lifetime, unproven at scale | LM26 demo machine | 2030s |
| Levitated Dipole Dipole |
Single levitating magnet confines plasma around it (like Jupiter) | OpenStar | Simplest design, cheapest prototype ($10M), inherently stable | Very early stage, single company, unproven energy extraction | First plasma demo | TBD |
| Sheared-Flow Z-Pinch Z-Pinch |
Electric current creates self-compressing magnetic field | Zap Energy | No external magnets at all, very compact, cheapest possible | Classic z-pinch instabilities, sheared-flow is new, short pulses | Gigapascal pressures | TBD |
General Fusion's Magnetized Target Fusion approach and the landmark SVAC SPAC transaction.
Via SPAC merger with Spring Valley Acquisition Corp. III (NASDAQ: SVAC). F-4 filed February 25, 2026. Expected close mid-2026. Proceeds fund the LM26 demonstration machine to validate Magnetized Target Fusion at commercial scale.
| Company | Vehicle | Valuation | Technology | Stage | Year |
|---|---|---|---|---|---|
| General Fusion (GFUZ) | SVAC SPAC | ~$1B | Magnetized Target Fusion | LM26 demo | 2026 |
| TAE Technologies | Trump Media merger | $3.1B | FRC + Beam Injection | 50 MWe plant | 2025 |
| NuScale Power (SMR) | SPAC | $1.9B | Small Modular Reactor (fission) | NRC certified | 2022 |
First-mover public fusion stock creates "category-defining" premium. MTF is the simplest, most capital-efficient path — liquid metal compression avoids extreme conditions. $335M funds LM26 milestones that de-risk technology for next-stage capital. 20-year track record of fusion results. CEO Greg Twinney is credible. Oversubscribed PIPE signals institutional confidence. If LM26 demonstrates net energy gain, GFUZ could 10×+.
SPAC track record is awful — most fusion/energy SPACs have destroyed shareholder value (NuScale -80% from peak). MTF is the least-validated approach among major contenders. Piston synchronization at commercial scale is unproven. Redemption risk: if SVAC holders redeem, proceeds shrink dramatically. Pre-revenue for years. Competing approaches (CFS tokamak, Helion FRC) are further along technically. SPAC structure creates dilution overhang.
On February 26, 2026, the NRC proposed the first dedicated regulatory framework for commercial fusion — a landmark moment for the industry.
The NRC chose to regulate fusion under its byproduct material framework (10 CFR Part 30) rather than as nuclear power reactors (Parts 50/52). This is the most developer-friendly outcome possible — it means fusion gets performance-based, technology-inclusive, risk-informed regulation instead of the prescriptive reactor licensing that has made fission so expensive.
A machine capable of (1) transforming atomic nuclei through fusion into different elements/isotopes/particles and (2) directly capturing and using resultant products including particles, heat, or electromagnetic radiation.
Updates definition to include radioactive material generated by, or made radioactive by use of, a fusion machine. Resolves jurisdictional uncertainty required by ADVANCE Act of 2024.
Must address: design, radiation protection, materials handling, organizational safety, training, maintenance, and accounting. Room for alternative approaches.
Tritium fuel, neutron activation products in structures, activated dust and components. These align with existing materials regulatory practices — no new paradigm needed.
The NRC's decision to regulate fusion as materials, not reactors, removes the single biggest regulatory overhang on the sector. Fusion developers can now plan licensing with confidence. This makes the 2027–2030 commercialization window achievable from a regulatory standpoint. Investors should view this as de-risking the entire sector.
Critical past achievements and upcoming catalysts. Items marked NEW are since v1.
Three structural shifts since our v1 thesis: GFUZ creates a tradeable vehicle, the NRC framework removes regulatory overhang, and stellarators emerged as a serious contender.
Everything we wrote in v1 is now more true. The NRC chose the lightest possible regulatory framework — fusion-as-materials, not reactors. GFUZ gives investors a first tradeable vehicle for pure fusion exposure. CFS confirmed SPARC 2027 — the single biggest near-term catalyst. Proxima's €1.7B Alpha MOU shows European industrial capital entering the race. OpenStar proved levitated dipole works with a $10M prototype. The capital base has expanded from $4.5B to $8B+. Multiple approaches are converging on physics milestones simultaneously. The "iPhone 2007" analogy from v1 is becoming "iPhone 2008" — we're past the demo, approaching the app store.
No private company has achieved net energy gain — this hasn't changed. SPAC deals in energy have a terrible track record (NuScale -80%, QuantumScape -90%). Solar + wind + storage costs continue to plummet. The engineering gap between "physics milestone" and "commercial reactor" is enormous. Most fusion capital is still going to pre-revenue companies with 5-10 year timelines. The NRC framework, while positive, still requires licensing — it's easier, not free. ITER's infinite delays remind us that fusion timelines are notoriously optimistic. Stellarator manufacturing at scale is unproven. Correlation risk: if any major company fails publicly (SPARC misses, GFUZ burns cash), it could damage the entire sector's credibility.
| Strategy | Vehicle | Risk | Reward | Catalyst |
|---|---|---|---|---|
| Direct Fusion Exposure | GFUZ (mid-2026) | 🔴 High | 🟢 Very High | SPAC close, LM26 milestones |
| Indirect: Nuclear Supply Chain | BWX Technologies (BWXT), Brookfield Renewable (BEP) | 🟡 Medium | 🟢 High | Regulatory clarity, order flow |
| Stellarator Pick & Shovel | Advanced manufacturing / HTS suppliers | 🟡 Medium | 🟢 High | Proxima Alpha, Type One orders |
| Energy Transition ETF | QCLN, ICLN (broad clean energy) | 🟢 Low | 🟡 Moderate | Macro energy policy |
| Wait & Watch | Cash + monitor SPARC 2027 | 🟢 None | — | CFS SPARC first plasma |
Three structural improvements justify an upgrade: (1) NRC regulatory clarity removes the biggest non-technical risk; (2) GFUZ creates investability where there was none; (3) Multiple companies are hitting physics milestones simultaneously, reducing single-point-of-failure risk. Key watch: CFS SPARC first plasma (2027). If it works, fusion becomes a trillion-dollar inevitability. If not, patience into the 2030s.
Primary sources for all data and claims in Fusion Atlas v2.