In fusion, small atoms are squeezed together to create larger ones and release oodles of energy. It has long been a perennial technology of tomorrow, but recent advances are turning it into a real business. See also nuclear fusion for the Farnsworth fusor and Orbitron approaches. See ITER for the long-running international project.
In the most common approach, the squeezing is done by powerful magnetic fields inside a doughnut-shaped container called a tokamak filled with superhot plasma. The bigger the tokamak, and the stronger the magnetic field, the more power it can produce. Since their invention in the 1960s, tokamaks have been improving faster than the famously exponential Moore's Law for microprocessors. Only now are they on the cusp of generating more power than they use to create the reaction in a way that can be scaled.
Commonwealth Fusion Systems (CFS) and Helion, two hot fusion startups together worth some $10bn, have signed contracts to provide power to, respectively, Microsoft (from 2028) and Google and a big Italian client (from the early 2030s). TAE Technologies agreed a $6bn merger with Donald Trump's social-media firm in December 2025.
MIT's Plasma Science and Fusion Centre comprises 23,000 square metres of lab space in seven buildings with 250 researchers, staff and students. It was led from 2024 by Nuno Loureiro, a Portuguese plasma physicist who was killed in December 2025.
CFS credits the development of "high-temperature" superconductors, which operate at -200°C rather than -270°C and require far less power to cool. They also create stronger magnetic fields than alternative materials. Artificial intelligence helps control the fickle plasma inside tokamaks.
China's state-led fusion effort has become a fierce competitor to Western programmes, built on a three-pronged strategy: setting research priorities, providing vast funding and building an industrial supply chain for reactor components. Chinese engineering firms have particular expertise in metallic carpentry, magnetic coils and power-conversion components; ITER, the long-running international fusion project in southern France, uses Chinese-made parts.
In 2025 China created China Fusion Energy, a state-owned enterprise under its national nuclear company, to tie together research efforts. On January 15th 2026 a new Atomic Energy Law went into effect, driving investment by setting out regulations. In March 2026 the government included nuclear fusion in high-level economic blueprints, including the 15th Five Year Plan.
China is building the Burning Plasma Experimental Superconducting Tokamak (BEST) in Hefei, a research hub. Construction is on track to be completed by 2027. BEST is designed to fuse deuterium and tritium; it will initially rely on an external tritium supply but aims eventually to produce its own by lining the vessel with a lithium blanket that converts neutrons into tritium. BEST will be a test bed for the China Fusion Engineering Demo Reactor (CFEDR), expected to be operational by 2030, with grid-connected power stations to follow—a timeline at least a decade ahead of other governments' efforts.
On January 1st 2026 researchers at EAST (Experimental Advanced Superconducting Tokamak), one of BEST's predecessors, reported that they had increased plasma density to levels once thought impossible. Down the road from BEST, the Comprehensive Research Facility for Fusion Technology (CRAFT), nicknamed Kuafu, is developing materials, magnets and components for future devices, as well as high-precision robots that can carry heavy payloads under high temperatures to maintain reactors. Yannick Marandet, research director of France's National Centre for Scientific Research, has said that the big advantage the Chinese have is their willingness to "learn by doing", in contrast with Europeans who want to perfect technologies before construction.
Across the world 77 startups have raised $15bn to pursue fusion using technologies ranging from advanced tokamaks to laser-driven designs and stellarators. Some American firms claim they will supply energy to the grid by the early 2030s. In America, 42 startups have raised a total of $8bn to date; eight Chinese firms raised about $5bn much more quickly. NovaFusionX, a Chinese firm, raised $70m in April 2025, the largest first-funding round for a private fusion company in China. Startorus Fusion, a Tsinghua University spinoff betting on a spherical-shaped tokamak, raised double that in January 2026. Energy Singularity hopes to reach the same goal by building extremely strong magnets, while ENN, a Chinese conglomerate, is attempting to fuse hydrogen nuclei with those of boron.
The fusion supply chain has grown more robust. Thanks to reliable orders from CFS, its superconductor suppliers have increased production by a factor of 40 since the startup was founded in 2018. Few fusion companies are generating revenue and none is making a profit.
The root of all superstition is that men observe when a thing hits, but not when it misses.