Why do nuclear weapons expire?

in

This is the script for my video posted on September 2, 2026. Watch it on TikTok.

Most people don’t know that nuclear weapons go bad, like bananas getting overripe on your kitchen counter. It’s true: tritium decays by roughly 5.5% every year, and without it a nuclear warhead with a yield of 100 kilotons would yield only 0.3 kilotons. But why is this the case, and what impact does it have on geopolitics?

Tritium is an isotope of regular old hydrogen, which has just a single proton in its nucleus. Deuterium adds one neutron, and tritium, sometimes called hydrogen-3, has two. Chemically, it behaves just like hydrogen, even able to form a kind of water, but that overloaded nucleus is unstable.

Tritium consequently has a half-life of just over 12 years, meaning that in that time, half of the atoms in any sample have turned into something else. One of those extra neutrons flips into a proton, spits out an electron, and the atom becomes helium-3. No mass is lost; it just stops being tritium.

Tritium is extremely rare in nature, at about one atom per billion billion regular hydrogen atoms. In that environment, they’re produced when cosmic rays strike the upper atmosphere, but tritium can also be manufactured by irradiating lithium-6 with neutrons in a reactor.

They can also be harvested as a byproduct of nuclear reactors that use heavy water, where neutrons from the reaction hit the deuterium in the heavy water and merge with it. Tritium costs about 30000 dollars per gram, about the same as natural diamonds.

So why does tritium matter for nuclear weapons? Well, you can think of it kind of like nitrous oxide in a car’s engine. Tritium isn't the explosive itself, that's still plutonium or enriched uranium, but a few grams of deuterium-tritium gas can multiply the yield by roughly a factor of ten. As a fission weapon’s chain reaction begins, the gas can fuse and release a flood of high-energy neutrons that cause far more of the fissile material to split before the pit can blow itself apart.

This allows higher yields, but it also means that less fissile material needs to be used to hit the same yield, allowing for smaller and lighter warheads and selectable yield designs.

Because tritium decays so quickly and produces helium-3 as a byproduct, which absorbs the very neutrons that boosting depends on, reservoirs on nuclear weapons have to be periodically removed, stripped of helium-3, and installed with fresh tritium. In practice these reservoirs get swapped out every five to ten years, and a warhead that misses its refill becomes unreliable within a decade or so.

As you can imagine, this makes a stable tritium supply extremely important for nuclear deterrence. For decades, nearly all US tritium came from the Savannah River Site in South Carolina, but all reactors there were shut down by 1988 and for roughly 15 years the US produced no new tritium, sustaining its arsenal instead by recycling gas recovered from dismantled nuclear warheads.

In the late 1990s though, the US Department of Energy decided to irradiate rods containing lithium-6 at civilian nuclear reactors, mainly the Watts Bar reactor in Tennessee, which coincidentally my grandfather used to work at. Irradiated rods would get shipped off to an extraction facility at the Savannah River Site after exposure.

But this was highly controversial. The US had a long-standing policy against dual-use, strictly separating civilian and military nuclear programs. Using a civilian electricity-generating reactor to make weapons material blurred this line.

Despite this, the US decided in 2024 to raise the limit on lithium-6 irradiation, achieving a record 13 tritium extractions in a nine month period as of January 2026. The US is increasing its tritium production capacity so that it’s able to expand, hedge, or otherwise adjust its nuclear posture amid competition with Russia and China.

See, in February 2026, the New START Treaty expired, meaning that there are no longer any binding limits on US and Russian strategic arsenals. This has prompted fears of a nuclear arms race, and though the Congressional Budget Office estimates nearly a trillion dollars in nuclear spending through 2034, most of that replaces aging Cold War systems rather than adding warheads.

Actually growing the arsenal runs into hard bottlenecks, and tritium is one of them, alongside plutonium pit production and the new missiles and submarines that keep slipping behind schedule.

Russia runs dedicated tritium reactors at Mayak. China is assessed, though not confirmed, to be pulling tritium out of its civilian heavy water reactors at Qinshan. France shut down its dedicated tritium reactors in 2009, and after sustaining itself on reserves for fifteen years it adopted the US model in 2024, irradiating lithium in a civilian power plant.

The UK has no tritium production at all and depends on US shipments under a 1958 agreement, so while the UK holds its own inventory, over time US cooperation is what keeps the British deterrent viable. Canada produces and stores plenty of tritium as a byproduct of nuclear power, but Canadian law bars exporting it for weapons, and most of it is spoken for by fusion research anyway so it can’t share with the UK.

And these are just the strategic nuclear powers. India, Pakistan, Israel, and North Korea are all assessed to produce tritium too, though for these countries we're working from intelligence estimates and open-source modeling, not confirmed numbers.

Tritium’s half-life makes it theoretically a very powerful lever for arms control, because plutonium and highly enriched uranium last effectively forever. Halting tritium production means that arsenals will decay, and controls on tritium have been proposed, but they’ve never been adopted because nuclear powers have no interest in degrading their own deterrent.

But tritium would also be extremely important if commercial fusion reactors came online. It’s estimated that a single 1 gigawatt fusion reactor could require about 55kg of tritium to operate for a year when the entire world only has around 20kg in total.

Fusion plants are hypothetically meant to create their own tritium during operations, but this is obviously unproven at a commercial scale and there’s a chicken-and-egg dynamic to it because you need tritium to start the reactor to begin with.

That tension between civilian and military nuclear programs comes up again here, because if fusion reaches a commercial scale and everyone needs tritium, whoever makes it will be sitting on a material that works just as well in a warhead as in a power plant, and there is currently no international safeguard regime for it.

China, which is adding roughly a hundred warheads a year, and Russia, which is modernizing at roughly flat numbers, both run their civilian nuclear programs with no separation from the military side. And to be fair, the US, UK, and France are all ramping up tritium capacity right now too. No treaty has ever limited tritium, and New START's expiration doesn't change that. What it does change is that for the first time in over fifty years, there is no accountability for warheads, and the one material every arsenal depends on is about to become the fuel that every fusion startup on Earth is fighting over.

Keep reading

I’m writing a book called Groundwork about the buried history beneath geopolitics. Join the waitlist to hear when it has a publisher, a cover, and a release date.