Nuclear Fusion: The Physics, the Machines, and the Engineering Gap
Fusion is the reaction that powers every star, and it is the only energy source that has been simultaneously twenty years away for seventy years and closer than it has ever been. Both statements are true, and the reason they are both true has more to do with materials science and tritium supply than with the physics of the reaction itself.
Why Fusion Releases Energy at All
Every atomic nucleus sits somewhere on a curve of binding energy per nucleon. Iron sits at the bottom of that curve, which is the most stable place to be. Nuclei lighter than iron can release energy by merging; nuclei heavier than iron release energy by splitting. That single curve explains why nuclear fission works with uranium and fusion works with hydrogen, and why nothing works with iron.
The reaction easiest to achieve on Earth combines deuterium and tritium, two heavy isotopes of hydrogen. They produce a helium nucleus and a fast neutron, and the products together weigh about 0.4 percent less than the reactants. Multiply that fraction by the speed of light squared and a gram of fuel carries roughly the energy of eight tonnes of oil.
The obstacle is electrostatic. Both nuclei are positively charged and repel each other, and the repulsion rises steeply as they approach. Overcoming it requires temperatures around 100 million degrees - nearly ten times the centre of the Sun, because the Sun compensates with a gravitational pressure no machine can match and takes billions of years to burn its fuel. A power plant does not have billions of years.
Two Ways to Hold a Plasma
At those temperatures matter is a plasma: nuclei and electrons moving separately, and no material container can touch it without being destroyed and cooling the plasma instantly. Two strategies exist, and they differ in how long they need to work.
Magnetic confinement uses the fact that charged particles spiral along magnetic field lines. A tokamak bends those lines into a torus so the particles circulate without hitting a wall. ITER, under construction in southern France as a collaboration of 35 countries, is the largest example - a 23,000-tonne machine designed to produce 500 megawatts of fusion power from 50 megawatts of heating. Stellarators such as Wendelstein 7-X in Germany achieve the same twist with intricately shaped coils rather than a plasma current, which is harder to build and easier to run continuously.
Inertial confinement takes the opposite approach: compress a fuel pellet so violently that it fuses before it has time to fly apart. The National Ignition Facility in California focuses 192 laser beams onto a capsule a few millimetres across, and the whole event lasts nanoseconds. The technique was developed as much for weapons physics as for energy, which shapes what the facility is optimised for.
Neither approach is a sideline. Both have absorbed decades of work, and the engineering problems they face are different enough that it remains genuinely unclear which will reach a power plant first.
What Happened in December 2022
On 5 December 2022 the NIF delivered 2.05 megajoules of laser energy to a fuel capsule and the capsule released 3.15 megajoules. For the first time, a fusion reaction in a laboratory gave back more than was put into it at that boundary. The result is genuine and was reproduced in later shots with higher yields.
The boundary matters. The lasers themselves drew several hundred megajoules from the grid to deliver those 2.05, because flashlamp-pumped glass lasers are inefficient by design - NIF was built for precision, not economy. Counted from the wall socket, the shot consumed on the order of a hundred times more energy than it released.
This is not a criticism of the result; it is what the result was designed to measure. Scientific breakeven and engineering breakeven are separate thresholds, and the field has always distinguished them. The 2022 shot crossed the first. No machine has yet crossed the second.
The Problems That Remain
Tritium is the first. It has a half-life of roughly twelve years, so it does not occur naturally in useful quantities, and the world's civilian stock amounts to a few tens of kilograms, largely a by-product of Canadian heavy-water reactors. A fusion plant must breed its own tritium by capturing its neutrons in a lithium blanket. The chemistry works in principle and has never been demonstrated at plant scale in a closed cycle.
Materials are the second. The neutrons carrying most of the energy are far more energetic than those in a fission reactor, and they displace atoms in the structural steel they pass through. Over years of operation this swells and embrittles the vessel. Candidate alloys exist; a facility capable of testing them under representative conditions is still being built.
Continuity is the third. A tokamak plasma is typically pulsed, and a laser facility fires a few times a day, while a power plant must run for months. Closing the gap between a physics experiment and a machine that pays for its own upkeep is where most of the remaining work lies - and it is the same distinction that separates a laboratory result from a product in any energy technology, including the ones still in research.
Private companies have entered the field in numbers since around 2020, pursuing smaller devices with high-temperature superconducting magnets that were not available when ITER was designed. Whether that shortens the timeline is a live question rather than a settled one.
Frequently asked questions
Is fusion the same as the reaction in a nuclear power plant today?
No. Today's plants use fission, which splits heavy nuclei such as uranium-235. Fusion merges light nuclei. They sit on opposite sides of the binding-energy curve, use different fuels and produce different waste - fusion leaves no long-lived fission products, though the reactor structure itself becomes activated by neutrons.
Did fusion achieve net energy gain in 2022?
At the fuel capsule, yes: 3.15 megajoules out for 2.05 in. Measured from the electricity grid, no - the lasers consumed roughly a hundred times the energy the reaction released. Both numbers are correct; they measure different boundaries, and the distinction is standard in the field.
Why does fusion need higher temperatures than the Sun's core?
The Sun confines its plasma with immense gravitational pressure and is in no hurry - a given proton may wait billions of years to react. A power plant needs reactions now, at pressures a machine can produce, so it compensates with temperature: roughly 100 million degrees against the Sun's 15 million.
Is fusion radioactive?
The fuel cycle involves tritium, which is radioactive but short-lived and low-energy. The reaction produces no long-lived fission products. The reactor structure does become radioactive through neutron activation, and managing that material is part of the design - the timescales involved are typically decades rather than the millennia associated with fission waste.
When will fusion supply electricity?
ITER is designed to demonstrate the physics at power-plant scale rather than to generate electricity; a demonstration plant would follow. Published national roadmaps point to the second half of this century, and several private companies target the 2030s. These are targets, not schedules, and the tritium and materials questions above are why.
