For decades, humans have sought to harness the power of the stars to generate electricity here on Earth. And for nearly as long, achieving that goal always seemed just a decade away.
Now, a slew of startups are closer than ever before and rushing to build fusion reactors capable of putting power on the grid.
Fusion startups have drawn more than $10 billion in investment, with more than a dozen raising over $100 million. Many large funding rounds have closed in the last year, with investors drawn to the industry as energy demand from data centers ramps up and as fusion startups draw closer to the finish line.
At its core, fusion power seeks to use the energy released from the fusing of atoms to generate electricity. Humans have known how to fuse atoms for decades, from the hydrogen bomb — an example of uncontrolled nuclear fusion — to any of the myriad fusion devices built in labs around the world. Experimental fusion devices have been able to control nuclear fusion, and one has been able to generate more energy than was required to spark the reaction.
But none of them have been able to produce enough of a surplus to make a power plant possible.
To solve that problem, fusion startups are trying a number of different approaches. Experts have varying opinions on which have the best chance of success, though the industry is still in its infancy, so nothing is guaranteed.
Here is a brief overview of the main approaches to fusion power.
Techcrunch event
San Francisco, CA
|
October 13-15, 2026
Magnetic confinement
Magnetic confinement is one of the most widely used techniques, using strong magnetic fields to confine plasma, the soup of superheated particles that’s at the heart of a fusion device.
The magnets must be tremendously powerful. Commonwealth Fusion Systems (CFS), for example, is assembling magnets that can generate 20 tesla magnetic fields, which is about 13 times stronger than a typical MRI machine. To handle the amount of electricity required, the magnets are made out of high-temperature superconductors, which still need to be cooled to –253˚ C (–423˚ F) using liquid helium.
CFS is currently building a demonstration device called Sparc on a much more accelerated timeline in Massachusetts. The company anticipates turning it on sometime in late 2026, and if all goes well, it will begin construction on Arc, its commercial-scale power plant, in Virginia in 2027 or 2028.
There are two main types of fusion devices that use magnetic confinement: tokamaks and stellarators.
Tokamaks were first theorized by Soviet scientists in the 1950s, and since then, they’ve been widely studied. Tokamaks come in two basic shapes — a doughnut with a D-shaped profile and a sphere with a small hole in the middle. The Joint European Torus (JET) and ITER are two notable experimental tokamaks; JET operated in the UK between 1983 and 2023, while ITER is expected to begin operations in France in the late 2030s.
UK-based Tokamak Energy is working on a spherical tokamak design. Its ST40 experimental machine is currently undergoing upgrades.
Stellarators are the other main type of magnetic confinement device. They’re similar to tokamaks in that they keep the plasma contained within a doughnut-like shape. But unlike tokamak’s geometric sides, stellarators twist and turn. The irregular shape is determined by modeling the plasma’s behavior and tailoring the magnetic field to work with its quirks rather than force it into a regular shape.
Wendelstein 7-X, a large stellarator with modular superconducting coils that is operated by the Max Planck Institute for Plasma Physics. has been operating in Germany since 2015. Several startups are also developing their own stellarators, including Proxima Fusion, Renaissance Fusion, Thea Energy, and Type One Energy.
Inertial confinement
The other main approach to fusion is known as inertial confinement, which compresses fuel pellets until the atoms within fuse.
Most inertial confinement designs use pulses of laser light to compress fuel pellets. Several laser beams fire at once, and their pulses of light converge on the fuel pellet from all angles simultaneously.
So far, inertial confinement is the only approach that has broken a milestone known as scientific breakeven, which is when the reaction releases more energy than it consumed. Those experiments have occurred at the National Ignition Facility (NIF) at the Lawrence Livermore National Laboratory in California. Notably, measurements to determine scientific breakeven do not include things like the electricity required to power the experimental facility.
Still, nearly a dozen startups see enough promise in inertial confinement that they’re designing reactors around it. Focused Energy, Inertia Enterprises, Marvel Fusion, and Xcimer are some notable examples using lasers.
There are two companies that aren’t using lasers, though: First Light Fusion, which proposes using pistons, and Pacific Fusion, which plans to use electromagnetic pulses instead of lasers.
More to come
Those are the two main approaches to fusion power, though they aren’t the only ones. Soon, we’ll add more details about alternative designs including magnetized target fusion, magnetic-electrostatic confinement, and muon-catalyzed fusion.
