A machine in a lab outside Daejeon just held plasma seven times hotter than the core of the actual sun for 102 seconds straight — and somehow that number is more impressive the more you understand how hard it is to do.
If you'd told a fusion physicist in 2020 that a Korean reactor would hold 100-million-degree plasma for nearly two full minutes by 2026, most would have called it optimistic. In June 2026, the Korea Institute of Fusion Energy (KFE) announced exactly that: KSTAR, nicknamed the "artificial sun," sustained a plasma temperature of over 100 million degrees Celsius for 102 seconds, more than double its own previous record of 48 seconds set just two years earlier. It sounds like a lab curiosity. It's actually one of the clearer signals yet that fusion power is inching from theory toward engineering reality.
15M°C
100M°C
Scale is illustrative, not literal — but KSTAR's plasma really does run roughly seven times hotter than the sun's own core.
The Machine That's Been Chasing the Sun Since 2008
KSTAR — the Korea Superconducting Tokamak Advanced Research device — has been running inside the Korea Institute of Fusion Energy's campus in Daejeon since its first plasma shot in 2008, after a project approved back in 1995. A tokamak works by using powerful superconducting magnets to trap superheated, electrically charged gas — plasma — inside a donut-shaped chamber, holding it away from the walls long enough for fusion reactions to theoretically occur. The device is designed as a testbed for ITER, the massive multinational fusion project under construction in France, which means every second KSTAR adds to its confinement record becomes usable data for the whole global fusion effort, not just Korea's own program.
What Actually Changed: Swapping Carbon for Tungsten
The jump from 48 seconds to 102 seconds wasn't a matter of just running the machine longer and hoping. KFE's engineering team replaced the reactor's carbon plasma-facing tiles with a fully tungsten-walled internal structure. Carbon has a habit of absorbing tritium, one of the fuel isotopes future fusion reactors will use, which becomes a real problem at commercial scale. Tungsten doesn't have that issue, and it also happens to be more thermally stable under sustained bombardment — which is exactly what let the plasma hold together for twice as long without the walls degrading the reaction.
"A green light for acquiring the core technologies required for the fusion DEMO reactor."
— Suk Jae Yoo, President, Korea Institute of Fusion EnergyThe Climb to 102 Seconds, Bar by Bar
Notice 2020's bar is short but the temperature was already over 100 million degrees — it just couldn't hold there for more than 20 seconds. The real story of the last six years isn't reaching the temperature at all; it's learning to keep the walls, magnets, and fuel stable long enough to stay there. That's the unglamorous engineering work that actually decides whether fusion ever reaches a power grid.
How KSTAR Stacks Up
Insider's Insight: the "hotter than the sun" framing gets thrown around a lot in fusion coverage, and it's technically accurate but a little misleading on its own — the sun doesn't need to run nearly as hot because gravity does most of the confinement work for it. KSTAR has no gravity to lean on, so it has to brute-force the same reaction using magnetic fields and much higher heat. That's the whole reason fusion research is this hard in the first place.
The Road to 300 Seconds
Current: 102 secTarget: 300 sec
KFE's publicly stated next target is 300 seconds of sustained plasma confinement at fusion-relevant temperatures — a benchmark researchers consider a meaningful step toward the kind of continuous operation a commercial "DEMO" reactor would eventually need. Nobody at KFE is claiming KSTAR itself will ever generate a single watt of usable electricity; it's a research device, not a power plant. But every second added to that confinement time gets folded into the design data for ITER and for Korea's own planned K-DEMO reactor, which is meant to be the country's first fusion plant that actually attempts grid-scale output.
Personal Take: what strikes me most about KSTAR isn't the headline temperature number, honestly — it's how incremental this whole field is by design. Fusion research has spent decades absorbing "always 30 years away" jokes, and deservedly so at times. But 70 seconds in 2016 to 102 seconds in 2026, achieved through something as unglamorous as changing a wall material from carbon to tungsten, is the kind of boring, compounding progress that eventually stops being a punchline.
📌 Key Takeaways
- In June 2026, KSTAR sustained plasma at over 100 million degrees Celsius for 102 seconds, more than double its 2024 record of 48 seconds.
- The jump came from replacing carbon plasma-facing tiles with tungsten, which resists tritium absorption and holds up better under sustained heat.
- KFE's next target is 300 seconds, a milestone tied directly to the design of ITER and Korea's own planned K-DEMO fusion reactor.
What does KSTAR stand for?
Korea Superconducting Tokamak Advanced Research — a magnetic fusion device operated by the Korea Institute of Fusion Energy in Daejeon.
Does KSTAR produce usable electricity?
No — it's a research reactor built to test confinement techniques for future power-generating fusion plants like ITER and Korea's planned K-DEMO.
Why is 100 million degrees necessary?
Without gravity to help confine the reaction the way the sun's mass does, fusion on Earth needs much higher temperatures to force atomic nuclei to fuse.
What is the next milestone for KSTAR?
Sustaining plasma at fusion temperatures for 300 seconds, a target the Korea Institute of Fusion Energy has set as its next major benchmark.
None of this means a fusion-powered Korea is around the corner — commercial fusion power is still widely expected to be decades out, even by the researchers most invested in it. But 102 seconds is 102 seconds longer than anyone managed before, and the path from 20 seconds in 2020 to here has been a straight, if slow, climb. Keep an eye on Daejeon; the next number on that scoreboard might come sooner than you'd expect.
Tell me what you think in the comments below.