The one-paragraph version
Fission splits a heavy atom (uranium-235) into lighter ones and releases energy. It's how every commercial nuclear plant on Earth works today. Fusion joins two light atoms (deuterium + tritium) into a heavier one (helium) and releases more energy per unit mass — it's how the Sun works, and it's what 40+ private companies are racing to commercialize. Fusion can't melt down, produces no long-lived waste, and runs on fuel you can extract from seawater. The catch: holding a plasma at 150 million °C long enough to break even is one of the hardest engineering problems ever attempted.
Side-by-side
| Dimension | Fusion | Fission |
|---|---|---|
| Underlying reaction | Light nuclei (deuterium + tritium) fuse into helium-4 + a 14.1 MeV neutron | A heavy nucleus (uranium-235, plutonium-239) splits into two lighter nuclei + 2–3 neutrons |
| Energy per reaction | ~17.6 MeV per D-T event (~3.5× more energy per unit mass than fission) | ~200 MeV per U-235 event |
| Fuel source | Deuterium from seawater (effectively unlimited); tritium bred from lithium-6 inside the reactor | Mined uranium ore, enriched to 3–5% U-235 for power reactors (or HEU/plutonium for advanced designs) |
| Fuel availability | Deuterium: ~33 g per m³ of seawater. Lithium-6: terrestrial reserves measured in centuries-to-millennia at fusion scale | Known uranium reserves ~100 years at current consumption; longer with breeder reactors or thorium |
| Long-lived radioactive waste | None from the reaction itself. Activated structural materials decay to background in ~100 years with low-activation alloys | Spent fuel contains plutonium, americium, neptunium with half-lives of 24,000+ years — geological storage required |
| Meltdown / runaway risk | Physically impossible — any disturbance cools the plasma and the reaction stops instantly. No chain reaction | Sustained chain reaction must be actively controlled; loss-of-coolant accidents can melt the core (TMI, Chernobyl, Fukushima) |
| Proliferation risk | Low — no fissile material produced or required. Tritium has weapons relevance but quantities are tightly tracked | High — enrichment infrastructure and spent fuel reprocessing are dual-use |
| Engineering maturity | Pre-commercial. ~$10B+ private capital deployed; first demonstration plants targeted late 2020s to early 2030s | Mature. ~440 reactors operating worldwide, ~10% of global electricity |
| Neutron flux | Intense 14 MeV neutron flux — drives materials damage and tritium breeding | Lower-energy neutron spectrum (~2 MeV average); embrittles reactor pressure vessel over decades |
The physics, briefly
Every atomic nucleus sits somewhere on the binding-energy curve. Iron-56 is the most tightly bound nucleus in the universe — the valley floor. Heavy atoms (uranium, plutonium) sit to the right of iron and release energy when they split downhill toward iron. Light atoms (hydrogen, helium) sit to the left and release energy when they fuse downhill toward iron from the other side. Both reactions liberate the same thing — nuclear binding energy — but the engineering required to get there is completely different.
Fuel
Fission needs uranium-235, which is 0.7% of natural uranium. Commercial reactors enrich that to 3–5%. Known reserves last roughly a century at current burn.
Fusion needs deuterium (one in every 6,400 hydrogen atoms in seawater — practically unlimited) and tritium, which doesn't exist naturally in useful quantities. Tritium is bred inside the reactor by bombarding lithium-6 with the fusion neutrons. Closing that breeding cycle is one of the hardest open problems in fusion engineering.
Waste
Fission's spent fuel contains plutonium, americium, and neptunium with half-lives of 24,000+ years. Long-term storage (Yucca Mountain, Onkalo) is required by the chemistry, not by policy.
Fusion produces only helium directly. The reactor's structural materials become radioactive from neutron bombardment, but with modern low-activation steels they decay to background radiation within about a century — meaning no geological repository is needed.
Safety
A fission reactor sustains a chain reaction that has to be actively kept stable. Lose the coolant and the residual decay heat melts the core — that's TMI, Chernobyl, and Fukushima.
A fusion reactor holds just a few grams of plasma at any moment. Disturb it — a power blip, a vacuum breach, a magnet quench — and the plasma cools below fusion temperature within milliseconds and the reaction stops. There is no chain reaction, no critical mass, no meltdown failure mode. This is the single biggest reason fusion regulation will look nothing like fission regulation.
So why isn't fusion everywhere yet?
Because fusion's easy parts (fuel, waste, safety) come at the cost of its hard part: confining a 150-million-°C plasma long enough, and dense enough, to produce more energy than the magnets and lasers consume. Fission solved that in 1942 with Fermi's Chicago Pile. Fusion has been "30 years away" for 70 years — but the last decade has changed the slope. NIF reached ignition in 2022. REBCO superconducting magnets put commercial-scale tokamaks inside the budget of a private company. Over $10B of private capital is now deployed across 40+ developers, and the first grid-connected pilot plants are targeted between 2028 and the mid-2030s.