China's Huanliu-3 reactor has shattered temperature records, entered the burning plasma phase, and reshaped the global fusion hierarchy.
In a cavernous laboratory in Chengdu, Sichuan Province, a doughnut-shaped machine the height of a three-story building is doing something no other device in China has done before: sustaining hydrogen plasma at temperatures more than seven times hotter than the core of the Sun. The Huanliu-3 tokamak - known internationally as the HL-3 and operated by the Southwestern Institute of Physics (SWIP) under China National Nuclear Corporation (CNNC) - has achieved a series of breakthroughs in 2024 and 2025 that have fundamentally altered the global fusion timeline. After decades of careful progression from seconds-long pulses to reactor-relevant plasma conditions, China's newest and most powerful tokamak has crossed into territory where the physics of burning plasma becomes real. What was once a matter of theoretical projections is now a matter of engineering execution.
China runs two complementary "artificial sun" programs, each optimized for a different frontier of fusion physics. In Hefei, the Experimental Advanced Superconducting Tokamak (EAST) - the world's first fully superconducting non-circular tokamak - specializes in duration. On January 20, 2025, EAST achieved a world record by sustaining high-confinement plasma at over 100 million degrees Celsius for 1,066 seconds - more than 17 uninterrupted minutes. It shattered its own previous record of 403 seconds set in 2023. Song Yuntao, director of the Institute of Plasma Physics at the Chinese Academy of Sciences, described the significance plainly: a fusion device must achieve stable plasma operation for thousands of seconds to enable continuous power generation in future plants.
In Chengdu, the HL-3 plays a different role. While EAST proves duration, HL-3 proves intensity. Its mission is to push plasma performance to reactor-grade conditions - extreme temperatures, high plasma current, and the kind of energy confinement efficiency that will eventually power ITER, the international fusion megaproject under construction in southern France, and its Chinese successor, the China Fusion Engineering Test Reactor (CFETR). The two programs are not rivals but interlockers: EAST generates the long-pulse physics data that global fusion programs need; HL-3 generates the high-performance plasma data that China's own reactor roadmap demands.
In 2025, the HL-3 crossed a milestone its scientists had been targeting for years: the "dual 100 million degrees" threshold. For the first time, the device achieved simultaneous ion temperatures of 117 million degrees Celsius and electron temperatures of 160 million degrees Celsius in a single plasma discharge. "Our experiment has achieved the 'dual 100 million degrees' milestone, along with a major leap in overall fusion performance. This means China's fusion research is entering the burning plasma phase," said Zhong Wulu, chief designer of HL-3, in an interview with Chinese state broadcaster CCTV.
The physics behind this achievement is precise. In standard tokamak operation, electron temperatures and ion temperatures are roughly equal because the two populations exchange energy rapidly. Achieving a state where ion temperatures significantly exceed electron temperatures - the so-called "hot ion mode" - is considerably more difficult. The HL-3 accomplished this with ion-to-electron temperature ratios exceeding 2:1, which drove the device's fusion triple product - the key metric combining plasma density, ion temperature, and energy confinement time - above 0.69 x 10^20 keV s/m3. This figure approaches the conditions required for a working deuterium-tritium fusion reactor.
The mechanism that enabled this was the formation of what plasma physicists call double transport barriers: a strong internal transport barrier (ITB) inside the plasma core combined with a weaker edge transport barrier (ETB). These barriers reduce the rate at which energy leaks from the hot plasma center outward toward the vessel walls. Sustaining both simultaneously in a high-current, high-temperature plasma requires precise magnetic field shaping and high-power neutral beam injection - both areas in which SWIP has invested heavily over the past decade.
The HL-3's technical specification places it firmly in the front rank of global tokamaks. It has a major radius of 1.78 meters, a plasma current capacity of 2.5 to 3 million amperes, and a toroidal magnetic field strength of 2.2 to 3 Tesla. Its plasma geometry is highly shaped - elongated and triangular - which improves stability and allows the device to store more energy under the same magnetic field. The combination of large scale, high field, and shaped geometry is what makes the HL-3 particularly suited for reactor-relevant physics that smaller or less-shaped devices cannot access.
In recent upgrade cycles, SWIP commissioned two self-developed plasma heating systems: a high-power electron cyclotron heating system and a 7 MW neutral beam injection system. The neutral beam injectors - which fire high-energy deuterium atoms into the plasma to heat and stir the ions - are central to achieving the hot ion mode that drove the temperature records. The NBI systems achieved a maximum beam current and accelerating voltage of 40 amperes at 46 kilovolts in their most recent configuration.
Diagnostics have advanced in parallel. The HL-3 carries the world's first triple-grating precision spectrometer system - a charge-exchange recombination spectroscopy instrument with twice the measurement accuracy of international counterparts. This system measures ion temperature and rotational velocity with a radial spatial resolution of 1.5 centimeters, enabling the kind of detailed plasma profile reconstruction that is essential for understanding and controlling internal transport barriers. A 60-channel Thomson scattering diagnostic measures electron density and temperature across the full plasma cross-section with millisecond resolution. These instruments have not only served the HL-3's own research program - some of the diagnostic technologies developed at SWIP have been incorporated into the International Tokamak Physics Activity (ITPA) joint experiment program, contributing to the global knowledge base for ITER operation.
In November 2024, HL-3 launched a new experimental campaign featuring a digital twin system developed by CNNC. The digital twin - described by Chinese media as a "super eye" - creates a real-time virtual replica of the device's physical state, providing precise monitoring of the vacuum chamber baking process and enabling faster, more accurate control responses during plasma discharges. Machine learning algorithms are now being integrated into the control system for real-time plasma shaping, instability suppression, and disruption prediction.
One of the more significant developments in HL-3's recent history is not what it has measured but who it has invited. In December 2023, CNNC signed a cooperation agreement with the ITER Organization, making the HL-3 an official ITER satellite device. Since then, the device has been opened to international researchers on a scale unprecedented for a Chinese national fusion facility. The spring 2025 experimental campaign attracted joint research teams from the United States, France, Japan, South Korea, Portugal, and Thailand.
Their work targeted problems directly relevant to ITER's future operation: how to sustain high-beta plasma without triggering disruptions; how to mix deuterium and tritium isotopes in ways that maintain confinement efficiency; and how to manage the intense heat exhaust that reactor-grade plasmas generate at the vessel boundary. A prototype diagnostic system for ITER's charge-exchange recombination spectroscopy - the same type used on HL-3 - completed its initial technical validation on the Chengdu device, confirming that HL-3 results will transfer directly to ITER measurement systems.
Liu Ye, director of SWIP, has framed the international opening in explicitly competitive terms: "The HL-3 team will continue to develop key technologies and delve into frontier fusion plasma physics. The robust support strongly positions China to conduct burning plasma experiments and construct fusion reactors in the near future." In March 2024, CNNC announced that 10 of its nuclear technology research facilities - including HL-3 - would be opened to the world for the first time, a move that positions China not just as a fusion competitor but as a fusion host for the global research community.
Every experiment on the HL-3 feeds into a single, tightly managed national objective: the construction and operation of the China Fusion Engineering Test Reactor (CFETR). The CFETR, planned for Hefei, is China's bridge between experimental fusion and demonstration-scale power generation. Its engineering design was completed in 2020, and construction is expected to begin in the early 2030s. In Phase 1, CFETR aims to generate 200 megawatts of fusion power and demonstrate tritium self-sufficiency - the ability to breed its own tritium fuel from lithium blankets. In Phase 2, it will scale toward 1 gigawatt and serve as the test platform for materials and systems that will eventually go into a full commercial fusion power station.
In that roadmap, the HL-3 is indispensable. Its plasma performance data - particularly its work on high-ion-temperature scenarios, double transport barriers, and advanced divertor configurations - are the empirical foundation on which CFETR's operating scenarios are being designed. The "snowflake" and "tripod" divertor geometries being tested on HL-3, which distribute heat loads across wider surfaces to protect the reactor wall, are direct candidates for CFETR's plasma-facing component design. Without the HL-3's experimental campaigns, CFETR would be building into an empirical vacuum.
China's fusion ambition, at its core, is an energy security argument. As Li Bo, a senior engineer at SWIP, has noted, a single liter of deuterium extracted from seawater releases the energy equivalent of hundreds of liters of gasoline when fused - and China's coastline holds an effectively unlimited supply. The country's 2049 target for commercial fusion power generation is not a marketing headline but a stated national timeline tied to the same national planning infrastructure that has driven China's solar, wind, and high-speed rail build-outs.
The skeptical view of China's fusion achievements is worth taking seriously. Most of HL-3's record-setting discharges last milliseconds to seconds, not the thousands of seconds needed for a commercial plant. The hot ion mode used to achieve the temperature records involves deliberately imbalancing the ion and electron temperatures - a regime that produces impressive numbers but is difficult to sustain as plasma density increases. Western fusion programs, particularly Commonwealth Fusion Systems with its SPARC tokamak, are pursuing a different route using high-temperature superconducting magnets to achieve reactor conditions in a smaller, potentially faster commercial path.
The counterargument is structural. China's program is not primarily a race for individual records - it is a sequenced development program with defined transition points. The HL-3 is not trying to be a power plant; it is trying to generate the empirical data that makes CFETR's design choices defensible. Each record in ion temperature, each successful test of an advanced divertor configuration, each machine learning integr…