Fusion — forcing light atomic nuclei together to release energy — has been "thirty years away" for sixty years. Something changed: private fusion companies have drawn billions in capital, demonstrated first-of-their-kind physics results, and signed power deals with the world's largest technology companies. This guide covers the technical approaches, what the milestones mean, the economics of reaching the grid, and how to diligence a fusion company.
Why the money showed up
Three forces converged. First, physics: in 2022 the National Ignition Facility achieved scientific ignition, and in April 2025 it recorded a gain factor around 4.13 — 8.6 megajoules of fusion energy from 2.08 megajoules of laser input. Second, high-temperature superconducting tape shrank tokamaks from national-lab scale to startup scale. Third, AI data centers need enormous firm, carbon-free power, and hyperscalers now sign contracts for electricity that doesn't exist yet.
The main technical approaches
There is no single "fusion technology" — just a portfolio of confinement strategies trading physics difficulty against engineering difficulty:
Tokamak. The most studied approach: a doughnut-shaped chamber where magnetic fields hold plasma hotter than the sun's core. Commonwealth Fusion Systems is building SPARC, a compact HTS-magnet tokamak in Devens, Massachusetts, expected online late 2026 or early 2027.
Stellarator. A twistier tokamak cousin with steadier plasma but complex magnets. Type One Energy and Germany's Proxima Fusion (backed by Google in 2026) lead the private bets.
Field-reversed configuration (FRC). Helion Energy accelerates two plasma rings toward each other in an hourglass chamber and harvests electricity directly from the magnetic coils — no steam turbine. In February 2026, Helion's Polaris prototype became the first privately built machine to demonstrate measurable deuterium-tritium fusion at 150 million degrees Celsius. Helion closed a $465 million Series G at roughly a $15.5 billion valuation.
Magnetized target fusion. General Fusion compresses a magnetized plasma with a collapsing liquid-metal liner. In October 2026 it announced a world-first on LM26: electron temperatures above 1 keV, over 12 million degrees Celsius, with UK Atomic Energy Authority collaboration. It trades on Nasdaq as GFUZ — the first publicly listed pure-play fusion company.
Pulsed magnetic confinement. Pacific Fusion crushes a fuel target with banks of pulsed-power modules. It raised a $900 million milestone-tranched Series A led by General Catalyst and is building in Albuquerque. Inertia Enterprises raised a $450 million Series A in early 2026 for a laser-driven approach.
What "net energy gain" actually means
Q is the ratio of fusion energy released to energy put in to heat the fuel; Q above 1 is scientific breakeven. But there are layers. Scientific Q counts only energy delivered to the fuel — NIF's 4.13 is scientific; the lasers consumed vastly more wall-plug electricity. Engineering Q counts all facility energy consumed; nobody has achieved engineering breakeven. The Lawson criterion combines temperature, density, and confinement time into the triple product — the real threshold for a self-sustaining reaction; General Fusion targets it by the end of 2028.
When a company announces a temperature record, ask about density and confinement time. Temperature alone doesn't make electricity.
The timeline economics: first-of-a-kind plants and the PPA era
Fusion's economics run through first-of-a-kind (FOAK) plants: expensive and bespoke, with the business case resting on de-risking the nth-of-a-kind fleet. In 2025–2026, customers began underwriting the FOAK step:
- Microsoft and Helion: a PPA targeting 50 MW from Helion's Orion plant in Malaga, Washington, aimed at 2028.
- Google and CFS: a 200 MW PPA from CFS's inaugural ARC plant in Chesterfield County, Virginia — a 400 MW facility expected on the grid in the early 2030s — with options on future plants and a deepened equity stake, plus a DeepMind plasma-simulation partnership.
- Eni and CFS: a second ARC-linked PPA reportedly worth over $1 billion.
These deals validate that creditworthy buyers will pay for fusion electrons and give startups contracted revenue for construction. First grid power from the leading designs is an early-to-mid-2030s event.
The regulatory path is clearing
The ADVANCE Act codified separate U.S. treatment for fusion, and in February 2026 the NRC proposed regulating fusion machines under the byproduct-materials framework rather than as fission reactors. A lighter licensing path shortens the road from demonstration to deployment.
How to diligence a fusion company: an investor's checklist
- Triple-product trajectory, not headlines. Demand temperature, density, and confinement time together — and the trend across generations.
- Hardware cadence. Fusion is won by teams that build, learn, and rebuild. Ask how many machines the team has built and what each proved.
- Capital intensity per milestone. Compare dollars raised against physics delivered. Milestone-tranched rounds like Pacific Fusion's $900 million Series A align incentives.
- Credible offtake. A signed PPA with a creditworthy counterparty is the strongest commercial signal. Distinguish binding deals from MOUs.
- Regulatory and siting reality. Where will the first plant be built, under which framework, and with what grid-interconnection plan?
- Fuel and supply chain. D-T machines need rare, regulated tritium; HTS designs depend on superconducting tape supply.
- Path to nth-of-a-kind cost. How does the tenth plant cost a fraction of the first?
- Financing optionality. First plants will need project finance, not just venture equity.
Fusion is no longer a science project funded by hope — it is an infrastructure asset class funded by milestones, magnets, and megawatts under contract. Read the physics like an engineer and the contracts like a banker. Angeled.com connects founders with investors.
Read Next
Decoupling Reaches Seed VC: The MiraclePlus Sanctions and What They Mean
Nauta Closes €60M First Close for Fund VI to Back European B2B Software
Nvidia's French Startup Investments: Equity Stakes in Doctolib and Genesis AI Anchor Its European AI Play
Solid-State Batteries: The Commercialization Timeline Investors Should Actually Watch
The Virtual Power Plant Wave: Why Energy Just Became a Top VC Category
Where India's Consumer and Fintech VC Is Betting in Late 2026




