Part 2 of 4: What Is Burning in There — and Why Not Already?

Series: Part 1 of 4: Humanity’s Most Expensive Mistake · Part 3 of 4: The Parts of a Revolution — and Who Holds Them Back

Part 1 asked why a possible energy source with global reach receives so little attention. Now we need to look at what actually happens inside a fusion reactor.

The short version is this: take hydrogen, heat it beyond one hundred million degrees, and hold it away from the walls with magnetic fields. It sounds simple. It is about as simple as describing an airport by saying that people get on a plane and later get off somewhere else.

Fission and fusion are not the same

Fission splits heavy atomic nuclei such as uranium. Energy is released, along with radioactive fission products, and the chain reaction must be controlled.

Fusion joins light nuclei. The fuel considered for power plants is mainly deuterium and tritium, two heavy isotopes of hydrogen. The reaction produces helium and a fast neutron. The Sun uses a different fusion pathway, but the basic principle is the same: energy is released when light nuclei combine.

The differences from fission are substantial:

Deuterium is found in seawater. Tritium is scarce and would have to be produced from lithium inside the reactor.
There is no self-sustaining chain reaction. If temperature, density, or confinement fall below the required conditions, the plasma cools and fusion stops.
There are no long-lived fission products of the kind associated with conventional fission reactors. Reactor components will still be activated by neutrons, and tritium remains radioactive. “No problem” would be as inaccurate as “the same as fission.”
A fusion power plant is not a source of material for a conventional nuclear weapon.

The ITER safety overview is useful here precisely because it does not list advantages without mentioning the open questions. The point is not that fusion removes every danger. The point is that it has a different risk structure.

What happens inside the reactor

A mixture of deuterium and tritium becomes plasma: an electrically conductive gas whose atoms have lost their electrons. At temperatures above 100 million degrees, the particles move fast enough to overcome the electrical repulsion between their nuclei.

Fusion creates a helium nucleus, which remains in the plasma and initially helps heat it. The neutron carries most of the energy outward. It hits the reactor wall, deposits its kinetic energy as heat, and turns the fusion plant into a very complicated heat engine. At the end of the chain, turbine and generator are familiar technologies. The conditions in which they receive their heat are not.

At 100 million degrees, any known material would vaporise. The plasma therefore cannot touch the wall. Magnetic fields are the solution.

Tokamak or stellarator?

The tokamak: the established route

The tokamak is ring-shaped and has been the workhorse of fusion research since the 1960s. ITER in southern France is a tokamak. The geometry is comparatively simple, and decades of research have created a large base of experience.

The tokamak needs a strong current in the plasma to produce part of the magnetic field. That current cannot easily be maintained indefinitely, so operation is typically pulsed. The plasma current can also become unstable and collapse in a disruption, transferring plasma energy to the wall very quickly. This is one of the central operating questions, not a footnote.

The stellarator: complicated to manufacture

The stellarator generates the required magnetic geometry entirely with external coils. They are three-dimensionally twisted and not identical. Wendelstein 7-X has 50 superconducting, non-planar magnetic coils.

That makes construction difficult. It may make operation easier: there is no large plasma current, no tokamak disruption in the same sense, and the possibility of continuous operation. A plant that supplies heat evenly is easier to integrate with the conventional power-plant side than one that repeatedly starts and stops.

The price is geometry. Each coil has to be right, manufacturing tolerances are tight, and the whole machine becomes a three-dimensional puzzle. Mastering it does not automatically create a monopoly. It does create knowledge that cannot be copied overnight.

What Wendelstein 7-X actually proves

Wendelstein 7-X is not a power plant and is not meant to become one. The Greifswald facility tests whether the stellarator principle is suitable for long plasma discharges and, eventually, a power-plant environment.

In February 2023, the plasma ran for eight minutes and reached an energy turnover of 1.3 gigajoules. In May 2025, Wendelstein 7-X held the triple product at record level for 43 seconds. An IPP report gives about 30 million degrees and 1.8 gigajoules. A pellet injector fed frozen hydrogen pellets into the plasma, while a water-cooled divertor removed heat.

These are important results. They do not show that the machine already produces net energy, operates a breeding blanket, or supplies electricity to the grid for years. The progress is that systems relevant to later steady-state operation are working together under demanding conditions.

The next step is not “switch on the power plant”. It is longer, hotter, repeatable operation with components that can survive the environment.

The six building blocks of a power plant

A hot plasma is not a power plant. At least six technical tasks stand between the two.

1. Superconducting magnets. Wendelstein 7-X uses low-temperature superconductors. Future concepts increasingly consider high-temperature superconductors, or HTS tapes. They can produce stronger fields and enable smaller reactors. The technology works, but it remains expensive and is not available in the industrial quantities required. ARPA-E gives current REBCO prices in the order of 300 dollars per kiloampere-metre and a long-term target near ten dollars.

2. The breeding blanket. Fusion neutrons must not only become heat. They should also turn lithium into new tritium. This blanket of breeding material, coolant, and structure is essential to a closed fuel cycle.

3. Neutron-resistant materials. 14-MeV neutrons alter materials at the atomic level. Steel can embrittle, swell, or creep; tungsten tolerates high temperatures but is not invulnerable. The important question is what materials do after years of real fusion-neutron exposure.

4. The divertor. It removes helium ash and impurities from the plasma and must handle extreme heat loads. Wendelstein 7-X’s island-divertor concept is among the more advanced components.

5. Heat extraction. Heat from the blanket, divertor, and other components has to enter a stable cycle. Steam, turbines, and generators are established technologies. Their integration into a fusion reactor is not.

6. Tritium supply. Tritium is radioactive, decays, and is scarce. The ITER data show why breeding cannot be treated as a later detail. New reactors need outside supplies for startup; during operation, the plant must breed more tritium than it consumes.

The tritium problem

Deuterium is the easy part. It occurs in seawater and is available in sufficient quantities. Tritium is the difficult part: it occurs naturally only in traces, decays with a half-life of 12.3 years, and cannot simply be stored forever.

A fusion plant must produce its own tritium. Neutrons hit lithium and create new tritium. Elegant in principle. In practice, the blanket must produce more than the plasma and plant consume. It also needs reserves for maintenance, losses, and the startup of later reactors.

That cycle has not yet been demonstrated in a commercial fusion reactor. This is where “the physics works” is not enough. Physics permits the concept. The full engineering chain at power-plant scale still has to be shown.

Proxima Fusion and the German ambition

In February 2026, Proxima Fusion, RWE, the Free State of Bavaria, and the Max Planck Institute for Plasma Physics agreed to cooperate on a commercial stellarator concept. Proxima plans an initial demonstrator called Alpha and a later plant called Stellaris. Bavaria offered up to 400 million euros. The agreement is an interesting industrial and political signal.

It is not yet proof of a reliable timetable. HTS magnets, blanket, materials qualification, and tritium supply must work together, not only separately. That difference is sometimes made smaller in press releases than it is in engineering.

The sober conclusion is this: the stellarator has a plausible advantage for continuous operation. It also pays a high price in construction and manufacturing. Wendelstein 7-X has shown that the central idea deserves to be taken seriously. It has not shown that the open questions have disappeared.

That is why fusion is not an academic game. The remaining tasks are large, expensive, and concrete. But a named engineering problem is easier to address politically than a myth about a technology that is always somewhere in the future.