Four hundred yuan per meter down to about one hundred. That single number — the cost China cut from a strand of high-temperature superconducting tape — tells you more about the future of fusion than any plasma temperature.
Because a fusion reactor eats that wire by the kilometer. When the unit price of the wire drops, as one Korean fusion engineer put it, it drops the cost of the whole machine. China didn't just build a magnet. It built the ability to build magnets cheaply, forever, without asking anyone.
Last week we covered how fusion gets licensed and financed. This week the story moves upstream, to the thing that makes national independence in fusion possible at all: a fully domestic supply chain for the hardware.
Start with the calendar, because the calendar is the argument.
On June 23, engineers at ITER — the seven-nation megaproject in southern France — finished stacking their central solenoid, the most powerful pulsed magnet ever assembled. Four days later, on June 27, the Institute of Plasma Physics in Hefei announced that its own 582-ton toroidal field coil had cleared full-parameter acceptance testing.
Same milestone. Two philosophies that cannot be reconciled.
The Hefei coil is a D-shaped monster: 21 meters long, 12 wide, 3.3 thick. It has 1.3 times the volume and three times the stored energy of ITER's comparable component. It is, by China's account, the largest fusion magnet ever built.
But the specs are not the point. The point is a single phrase that appears in every Chinese account of the achievement: 100% localized. Every strand of superconducting wire, every gram of cryogenic stainless steel, every insulation material and low-resistance joint — all produced inside China's own industrial base.
ITER, by contrast, is built through roughly nine-tenths in-kind contributions from its seven primary members (part of a 35-country endeavor). Seven primary members — China, the EU, India, Japan, South Korea, Russia and the United States — each contribute hardware and share the resulting intellectual property under a common framework.
That is the whole difference, and it is enormous.
ITER is fusion as a global commons. A small economy that ships a subassembly earns a seat at the table and a claim on the eventual licensing. The knowledge diffuses, unevenly but genuinely, across everyone who paid in.
Hefei is fusion as a sovereign industrial project — built fast, built in-house, answerable to no consortium and dependent on no foreign supplier. As one analysis of the divergence put it plainly: China's model offers no built-in mechanism for sharing.
ITER is fusion as a global commons. Hefei is fusion as a sovereign industrial project — answerable to no consortium and dependent on no foreign supplier.
So consider what happens if Hefei's 2030 target for fusion-generated electricity beats ITER's 2035 target for deuterium-deuterium operations.
The first commercially relevant fusion technology on the planet would arrive not through a multinational framework but through a single state's export relationships. Beijing would set every transfer term.
And we already know the shape of those terms, because we've watched China run this playbook before — in solar panels, in EV batteries, in fission reactors sold across Africa, Southeast Asia and Latin America through bilateral energy and infrastructure deals. Fusion would be the same bet aimed at a bigger target: the energy source underneath the AI economy.
That is the real story. Not who is faster. Who owns the thing when it works.
Now widen the lens, because the fracture runs deeper than two magnets.
On December 11, 2025, India's Institute for Plasma Research commissioned the country's first spherical tokamak and produced first plasma. Since then it has run 114 discharges — a genuine national milestone.
But the machine has a problem that reads almost as a metaphor for the whole field. When its magnetic system switches on, large unwanted electrical currents are induced in the metal walls of the vessel, and those currents produce magnetic effects that interfere with the plasma itself.
The researchers are trying to compensate by rearranging coils. A spherical tokamak, shaped less like a doughnut than a cored apple, is a bet on a different geometry entirely — incompatible with the ITER-scale toroidal machines Hefei is building.
Then there's the private track. Helion Energy just crossed 800 employees and opened a Seattle office, on top of five buildings in Everett and a plant site in Malaga where it hopes to stand up the world's first commercial fusion generator.
Helion has raised more than $1.5 billion and carries one of the most aggressive timelines in the industry: reportedly sufficient energy in two years. That clock isn't scientific ambition — it's contractual. The Malaga plant must be operational in two years to meet Helion's deal to power a nearby Microsoft data center.
And Helion's design isn't a tokamak at all. It smashes plasma together and harvests energy directly. It shares almost no engineering DNA with Hefei, with ITER, or with India's spherical machine.
Line them up. A Chinese state program building 582-ton low-temperature superconducting coils. A Korean effort chasing miniaturization with high-temperature superconductors. An Indian spherical tokamak fighting its own walls. An American startup racing a two-year corporate deadline with a device that looks like none of the others.
These are not competitors running the same race on the same track. They are incompatible bets, and the money, the headcount, and the record hardware are all fragmenting faster than any consensus about which design will actually work.
Which brings the sober caveat that has to close this out.
Every milestone here is real and every one is necessary. A magnet that passes full-parameter testing is not nothing. Cutting wire cost fourfold is not nothing. First plasma is not nothing. Eight hundred employees is not nothing.
But look at what a working machine actually demands. Wu Yu, the ASIPP researcher, spelled it out: a complete tokamak will need 16 of those 582-ton coils operating in unison, each carrying 100 kiloamperes and generating 6.5 tesla at the center.
Sixteen. In perfect synchrony, under extreme cryogenic conditions, in a machine that doesn't exist yet. Mass-producing and integrating them will test the outer limits of modern manufacturing — which is exactly the skill China is signaling it has, and exactly the thing no one has yet demonstrated in an operating power reactor.
Here is the thread we've been pulling since late August, reframed. The open question was whether any converging design could show net facility gain in hardware — more energy out of the building than went into it.
The answer, still, is that nobody has. Not Hefei. Not Helion. Not ITER, whose central solenoid won't drive full-power operations until the 2030s. Every claim of being "80% of the way" to commercial fusion is measured against a finish line no one has crossed.
That's the uncomfortable synthesis. The physics of net gain reportedly remains unproven on every track at once — while the geopolitics of ownership has already hardened into camps.
China has decided the decisive advantage won't be a shared equation. It will be a supply chain no one else controls. The 582-ton coil is less a scientific breakthrough than a manufacturing declaration, and the declaration is that the ITER dream of fusion-as-commons is quietly dying.
The race fragmented before the science converged. That's the part worth watching. Whoever lights the first sustained artificial sun will also own the wire, the steel, and the terms — and right now the country building all three at home isn't planning to share.