China Cracked a Fusion Barrier That Was Never Supposed to Break, and the Race to Build a Star on Earth Is Now Very Real
For decades, nuclear fusion has been the energy world’s most stubborn punchline, a technology that always seemed to be “the energy source of the future, and always will be.” The premise is almost absurdly ambitious: replicate the thermonuclear process that powers the sun inside a machine on Earth, and in doing so unlock near-limitless clean energy for the planet’s 8 billion inhabitants. But something has quietly shifted. The jokes are getting harder to make.
What’s changed isn’t any single breakthrough, it’s the accumulation of them. Plasma records are falling across three continents. A 23,000-tonne international reactor is inching toward completion. Artificial intelligence has entered the equation. And the private sector, which spent years watching from the sidelines, has opened the floodgates. When you layer all of that together, what emerges is a picture that even the most seasoned skeptics are finding difficult to dismiss.
Records Keep Falling, and the Science Is Getting Serious
The basic tool at the center of all this progress is the tokamak, a donut-shaped device first introduced in the 1950s that uses powerful magnetic fields to contain superheated hydrogen plasma, creating the conditions necessary for fusion to occur. Plasma, it’s worth noting, is the fourth state of matter, the point at which atoms shed their electrons entirely. Fusion primarily occurs in this state, at temperatures hovering around 100 million degrees Celsius.
In recent years, according to Popular Mechanics, tokamak reactors around the world have contained plasmas for hundreds of seconds or even a few dozen minutes at those blistering temperatures. France’s WEST reactor, China’s EAST reactor, and South Korea’s KSTAR reactor have each broken various records for plasma duration, all serving as testbed machines whose scientific findings are feeding directly into larger projects.
The most recent milestone came just this month, when China’s EAST reactor successfully surpassed the Greenwald limit, a theoretical plasma density barrier that describes how dense a plasma can become before it destabilizes. It is a line scientists long treated as a hard ceiling, and crossing it matters.
ITER and the Materials Problem Standing Between Fusion and Reality
All roads in the fusion world currently lead to one place: the International Thermonuclear Experimental Reactor, or ITER, under construction in southern France. A joint effort representing 35 countries, ITER is designed to push fusion science further than any machine in history, with the specific aim of demonstrating that a reactor can produce more energy than it consumes. The project has faced regrettable delays, but progress is visible. This past September, the sixth and final module of the central solenoid, described as the world’s most powerful magnet and the beating heart of the entire machine, arrived on site in France.
Still, even getting a reactor to produce net energy, the holy grail state known as “ignition”, is only part of the challenge. Doing so consistently, at scale, and cost-effectively is another matter entirely. One of the most stubborn obstacles is materials. Fusion reactors must withstand not just extreme heat, but also neutron flux, the description of neutron population and speed inside a reactor, which degrades materials over time.
To address this, MIT announced in June 2025 the establishment of theLaboratory for Materials in Nuclear Technologies, known as LMNT. “We need technologies today that will rapidly develop and test materials to support the commercialization of fusion energy,” said Zachary Hartwig, PhD, the lab’s head, in a press statement. “LMNT’s mission includes discovery science but seeks to go further, ultimately helping select the materials that will be used to build fusion power plants in the coming years.“
$9 Billion, Big Tech, and a World That’s Starting to Believe
Beyond the physics, perhaps the most striking signal of fusion’s shift from pipe dream to near-future reality is who is now putting money into it, and how much. According to the Kleinman Center for Energy Policy at the University of Pennsylvania, private sector investment in fusion totaled $1.5 billion between 2016 and 2020. From 2021 to 2025, that number jumped to nearly $9 billion.
Driving part of that surge is an unlikely catalyst: artificial intelligence. The need for corporations like Google, Microsoft, Amazon, and Meta to find energy sources capable of powering their expanding AI data centers has pushed big tech into fusion investment in a meaningful way. Some local utilities have gone further still, partnering directly with fusion startups to explore the technology. Governments around the world are also pouring in support, motivated by the urgent need to cut carbon emissions and avoid the worst consequences of climate change, including mass sea level rise.
AI is also reshaping the science itself. Researchers are developing models that can dynamically adjust magnetic fields inside plasma, fill missing fusion data with synthetic but reliable information, and identify safe zones within a reactor shielded from the plasma’s intense heat.
Significant hurdles, technical, scientific, and economic, remain. The sun is not giving up its fusion monopoly quietly. But, when measured against the 4.6-billion-year head start our star has had, what humans have managed to accomplish in just the past few years means fusion will arrive, in the grand scheme of things, in a blink of an eye.