The Machines · All six bottles
Same card for every machine in this atlas — no singled-out demo. Name, 3D, specs, company. Jump a method or scroll the whole catalog.
Magnetic · doughnut
The most studied bottle on Earth. One U.S. company is assembling a net-energy machine today.
Product · Commonwealth Fusion Systems
Fort Devens, Massachusetts. Compact HTS tokamak. Not a power plant — the proof. Designed for commercially relevant net energy (Q>1) with margin, targeting 2027. Eighteen D-shaped toroidal-field magnets. Fuel: deuterium–tritium. Plasma photographs pink (Balmer-alpha).
Kitty Hawk, not the airline.
On-axis. Prototype magnet 20 T.
Designed thermal output.
117 Hospital Rd, Devens.
Product · Commonwealth Fusion Systems
Fall Line Fusion Power Station, Chesterfield County, Virginia. The commercial follow-on: same HTS tokamak idea, sized for the grid. Molten-salt blanket catches neutrons, breeds tritium, makes steam. About 400 MW to the wire in company materials. Early 2030s target. Google contracted for ~200 MW.
Electrons on a wire.
Company design point.
Near Richmond, VA.
~half the first plant.
The builder
The U.S. private tokamak that is actually assembling a net-energy machine today. Magnet factory on campus in Fort Devens. Best-funded private fusion company (on the order of $4B in 2026 reporting). Tokamak Energy (UK) and Energy Singularity (China) are other compact-tokamak shops — not U.S. leads.
Magnetic · twist
Twisted bottles, U.S. steel. One mid-scale machine in Tennessee. One planar-coil design from Princeton.
Product · Type One Energy · Knoxville, Tennessee
Mid-scale stellarator to de-risk the path to a plant. Sited at the former Bull Run coal plant with TVA in East Tennessee. University of Wisconsin–Madison heritage — the co-founders built an optimized stellarator there in the 1990s. Infinity Two is the commercial follow-on in company materials (350 MW-class concepts, 2030s). DOE Milestone program. On the order of $170–200M disclosed.
De-risk the twist.
TVA coal plant reuse.
Plant-scale concept.
UW–Madison line.
The builder
University of Wisconsin–Madison stellarator line, now building in East Tennessee. Infinity One is the mid-scale machine; Infinity Two is the plant in company materials (350 MW-class, 2030s). DOE Milestone participant. On the order of $170–200M disclosed. Wendelstein 7-X in Germany is the lab that proved optimized stellarators work — Type One is the U.S. attempt to turn that into a plant.
Product · Thea Energy · Princeton region, New Jersey
PPPL spinout (formerly Princeton Stellarators). The product idea: a stellarator field from arrays of simpler planar coils instead of one nightmarish 3D winding. Preconceptual commercial plant design published 2026. DOE Milestone program. The machine is still on paper more than on a pad — that is the honest product status.
Plant concept first.
Arrays, not one twist.
Princeton spinout.
Milestone path.
The builder
PPPL spinout (formerly Princeton Stellarators), Princeton region, New Jersey. The company bet is manufacturing: planar coil arrays instead of one 3D winding. Preconceptual commercial plant design published 2026. DOE Milestone program. Honest status: a design house with a paper trail, not a pad poured yet.
Magnetic · compact
Two American companies, two FRC products, two electricity stories.
Product · Helion Energy · Everett, Washington
Seventh-generation pulsed FRC. Two plasmoids form, accelerate, and merge. 2026 coverage: 150 million °C-class plasmas and private D-T operation. Fuel path includes deuterium–helium-3. The commercial trick: harvest electricity from the changing magnetic field — no steam turbine in the pitch. Microsoft 50 MW PPA targeted 2028 from a plant near Malaga, WA. Treat 2028 as a target, not a fact.
Path to a 50 MW plant.
Also D-T in testing.
Inductive, not steam.
50 MW · 2028 target.
The builder
Everett, Washington. The U.S. colliding-FRC company, and one of the three best-funded private fusion shops. Direct electricity from a changing magnetic field is the pitch — no steam turbine in the story. Microsoft 50 MW offtake targeted 2028; treat that date as a target, not a fact. Nucor offtake also reported. Polaris is generation seven of the same idea.
Product · TAE Technologies · Foothill Ranch, California
Norman has run since 2017 — beam-driven FRC, the long-running private machine. Copernicus is the next device on the road to a plant. Destination fuel is aneutronic hydrogen-boron-11: fewer neutrons, different materials problem. Founded 1998. Investors have included Google and Chevron. ~$1.3–1.5B raised. Plant concepts in company materials sit in the early 2030s.
Operating FRC.
Scale-up machine.
Aneutronic path.
Longest private run.
The builder
Foothill Ranch, California. Founded 1998 — the longest-running private fusion company in the U.S. Beam-driven FRC: Norman has operated since 2017; Copernicus is next. Destination fuel is hydrogen-boron. Investors have included Google and Chevron. On the order of $1.3–1.5B raised. Da Vinci is the plant concept; early 2030s in company materials.
Magnetic · self-made
One U.S. company is commercializing sheared-flow Z-pinch. The machine is the column.
Product · Zap Energy · Everett, Washington
Sheared-flow-stabilized Z-pinch. Current down a plasma column makes the magnetic field that pinches it. No superconducting magnets, no lasers, no cryogenics in the confinement scheme. University of Washington spinout, 2017. DOE Milestone program. ~$338M disclosed. Same city as Helion — two bottles, two religions. Commercial plant concepts in company materials sit in the 2030s.
Path to a pilot plant.
Plasma is the magnet.
Washington State.
UW line.
The builder
University of Washington spinout, 2017. Everett, Washington — same city as Helion, different bottle. The U.S. commercial sheared-flow Z-pinch company. No superconducting magnets, no lasers, no cryogenics in the confinement scheme. DOE Milestone program. ~$338M disclosed. One American lead on this architecture.
Inertial · pulse
NIF proved ignition. These companies are trying to fire that shot like an engine.
Product · Pacific Fusion · Fremont, CA / Albuquerque, NM
Not a 192-laser cathedral. Pulsed electromagnetic power crushes the target. Founded 2023. ~$1B out of the gate. Team from Livermore and NNSA plus scaling talent from SpaceX/Tesla. Company target: net facility gain around 2030. Albuquerque facility groundbreaking in 2026 coverage. The product is the pulser plus the shot rate.
Electromagnetic, not laser.
Net facility gain.
Debut-scale raise.
New Mexico build.
The builder
Founded 2023. Fremont, California and Albuquerque, New Mexico. Pulsed-power inertial fusion — not a 192-laser NIF clone. Debut capital on the order of $1B. Team from Livermore and NNSA plus scaling talent from SpaceX/Tesla. General Catalyst among backers. Company target: net facility gain around 2030. The company is the pulser and the shot rate.
Product · Xcimer Energy · Denver, Colorado
Laser inertial confinement designed to pulse like a plant, not like NIF’s once-a-day shot. DOE Milestone program. On the order of $165M disclosed. Commercial concepts in public materials sit in the mid-2030s. The product is a laser architecture you can actually repeat.
Repeat-rate first.
Milestone path.
Colorado.
Plant concepts.
The builder
Denver, Colorado. Laser inertial confinement aimed at a plant cadence, not NIF’s once-a-day shot. DOE Milestone program. On the order of $165M disclosed. Mid-2030s commercial concepts in public materials. The company is the repeat-rate laser architecture.
Product · Inertia · Livermore, California
The NIF geometry — hohlraum, lasers, capsule — aimed at a commercial cadence. ~$450M Series A in 2026 reporting. Sitting next to the lab that proved ignition is the product story. Mid-2030s energy-production talk in industry tables.
Indirect drive.
Series A, 2026.
Next to NIF.
Industry tables.
The builder
Livermore, California — next to the lab that proved ignition. Indirect-drive laser ICF (hohlraum, lasers, capsule) aimed at a commercial pulse rate. ~$450M Series A in 2026 reporting. Mid-2030s energy-production talk in industry tables. Sitting beside NIF is the origin story.
Magnetic · open
An open bottle, reopened with high-temperature superconductors. One U.S. commercial lead.
Product · Realta Fusion · Madison, Wisconsin
Two magnetic corks, a cigar of plasma between them. University of Wisconsin–Madison spinout, 2022. HTS magnets raise the mirror ratio that killed 1980s programs. Aimed at industrial heat and power, not only a gigawatt grid plant. DOE Milestone program. ~$58M disclosed. 2026 reporting included a path toward a ~200 MWe plant concept with Madison Gas and Electric. Early-to-mid 2030s in industry tables.
Heat + power concept.
MGE path, reported.
The new cork.
UW–Madison.
The builder
University of Wisconsin–Madison spinout, 2022. Madison. The U.S. commercial tandem-mirror company. HTS magnets are the cork that 1980s mirrors did not have. Aimed at industrial heat as well as power — not only a gigawatt grid plant. DOE Milestone program. ~$58M disclosed. A ~200 MWe path with Madison Gas and Electric appeared in 2026 reporting; that is a concept, not a plant.