What if the most dangerous object in Europe is guarded by something that looks like a calculator you'd buy at RadioShack?
That's... a very specific question, Corn.
It's where Daniel's prompt ends up. Here's what he wrote in this week. A few episodes back he asked what a nuclear warhead actually looks like, thinking it was an obscure question, and it turns out there's a whole genre of YouTube material on it. Maintenance convoys, the physics package, all of it. And the thing that surprised him most is that the part that actually triggers the explosion is tiny. Then he says his second wrong assumption was the numbers. He'd guessed each nuclear state keeps a few of these in a highly secure facility. Turns out there are a lot more. And then the third thing he stumbled onto, which is the one he really wants us to dig into, is that the US distributes its arsenal across allied states. He says it makes abundant sense to push the envelope of deterrence out geographically, but it seems like it would require enormous trust. You're putting nuclear weapons in another country and hoping they use them responsibly.
Or at least adhere to your mutual understanding.
Right. So his actual questions are: how does this whole enterprise work exactly, and among the confirmed nuclear powers, how many of them choose this distributed architecture?
Three misconceptions, and the third one has the best answer.
Let's start with the warhead. Why is the physics package small, and why is the warhead not?
The physics package is the nuclear explosive package. That's the primary and the secondary, the actual assembly that produces the explosion. Everything else in a warhead is engineering built around it. The primary is a fission trigger, typically boosted fission, and the secondary is the fusion stage. Radiation from the primary's detonation is contained and used to compress and ignite the secondary. That's the Teller-Ulam staged radiation implosion design, and it's been the basic architecture since the early fifties.
So the primary is a bomb that exists to light a bigger bomb.
That's exactly what it is. And the package itself sits inside a uranium-238 reflector casing, with polystyrene foam around it. That's the structure. The nuclear part is small. The non-nuclear part is what makes it look like a warhead.
Give me the number that makes it land.
Fat Man's plutonium pit was nine point one centimeters in diameter. Three point six inches. The size of a softball. The bulk of Fat Man's girth was the implosion mechanism, concentric layers of uranium-238, aluminium, and high explosives. The nuclear core was the small thing in the middle.
Everyone pictures the bomb as the bomb. It's actually the packaging.
The W54 warhead, used in the Davy Crockett recoilless rifle, was twenty-eight centimeters across. Eleven inches. Put that next to Fat Man at a hundred and fifty centimeters and the shrink is the whole story of the second half of the twentieth century. The smallest nuclear weapon ever deployed was the W48 artillery shell.
An artillery shell.
The Swan device, which was the first two-point hollow-pit fusion-boosted implosion design and the prototype for basically everything that followed, was twenty-nine centimeters in diameter and fifty-eight centimeters long.
So the core is a softball, and the shell around it is the size of a person.
Modern pits contain three and a half to four and a half kilograms of plutonium and yield five to ten kilotons. That's fission of roughly half a kilogram of plutonium. Half a kilogram doing that much work.
That's the part that should be unsettling and somehow isn't, because we've all seen the diagrams.
The diagrams make it look like a machine. It is a machine. That's the point. The physics is small and the engineering is large.
But hold on. If the physics package is that small, why is a modern warhead still so heavy? Why can't you just put a softball in a suitcase?
You can, and people have worried about exactly that. But the casing isn't padding. It's the reflector, the radiation channel, the neutron initiator housing, the arming and fuzing electronics, the environmental control system, the reentry vehicle. A reentry vehicle has to survive atmospheric reentry at thousands of degrees. It has to arm at the right altitude. It has to not arm at the wrong altitude. Every one of those functions adds mass and volume, and none of them are the explosion.
So the bomb is the easy part and the delivery is the hard part.
The bomb was solved in the forties. The delivery problem is what kept engineers employed for seventy years.
Which is a nice segue, because the delivery problem is exactly where the numbers come in.
At the start of last year, nine states possessed approximately twelve thousand two hundred and forty-one nuclear weapons. Nine thousand six hundred and fourteen in military stockpiles. Three thousand nine hundred and twelve deployed with operational forces. And about two thousand one hundred of those on high operational alert on ballistic missiles.
Say the first number again.
Twelve thousand two hundred and forty-one.
Daniel guessed a few.
The nine are the US, Russia, the UK, France, China, India, Pakistan, North Korea, and Israel. The US and Russia together hold almost ninety percent of all warheads.
Break it down.
Russia around five thousand four hundred and fifty-nine. US around five thousand one hundred and seventy-seven. China around six hundred. France two hundred and ninety. UK two hundred and twenty-five. India one hundred and eighty. Pakistan one hundred and seventy. Israel ninety. North Korea fifty.
And that's where it gets interesting, because those numbers are contested.
The Federation of American Scientists estimates the US stockpile at around three thousand seven hundred and Russia at around four thousand three hundred and nine. That's not a rounding difference. That's a disagreement about what counts and what's knowable.
How do two organizations with good access disagree by fifteen hundred warheads?
Different definitions of stockpile, different assumptions about dismantlement rates, different treatment of weapons awaiting disassembly. And Russia doesn't publish. Nobody really publishes. The numbers are estimates built from treaty data, satellite imagery, and history.
So the honest answer to how many warheads exist is a range.
With a decent center of mass. SIPRI's twelve thousand is the best public figure. It's just not a fact in the way people want it to be.
What about the trend?
The overall number is declining, because the US and Russia are dismantling retired warheads. But SIPRI warns the dismantlement rate is slowing and will soon be outpaced by the rate at which new warheads enter global stockpiles. Their line is that the era of reductions since the end of the Cold War is coming to an end.
So the curve is flattening and about to turn.
China is the one to watch. SIPRI's director said China could reach a thousand warheads in seven or eight years. They're at six hundred now.
That's a build, not a modernization.
It's both. They're replacing older systems and adding new ones, and the silo fields in the west suggest the ceiling is higher than the current number.
And the silo fields are visible from space, which is the weird part. You can watch a country build a nuclear arsenal on Google Earth.
You can watch the construction. You can't watch what goes in them. That gap between what's observable and what's knowable is the entire discipline of nuclear estimation.
Okay. Now the third misconception, which is the one Daniel actually cares about. Nuclear weapons stay home.
The US stations nuclear weapons in Belgium, Germany, Italy, the Netherlands, Turkey, and the United Kingdom. As of last year, a hundred and twenty-five to a hundred and thirty tactical B61 bombs are believed deployed in Europe under nuclear sharing.
Give me the base-by-base.
Belgium, Kleine Brogel, ten to fifteen. Germany, Büchel, ten to fifteen. Italy, Aviano twenty to thirty and Ghedi ten to fifteen. Netherlands, Volkel, ten to fifteen. Turkey, Incirlik, twenty to thirty. UK, Lakenheath, twenty-five to thirty.
Lakenheath is the new one.
In July of last year, experts concluded twenty-five to thirty US nuclear weapons had been redeployed to RAF Lakenheath for the first time since two thousand eight. The evidence was telemetry from a Boeing C-17 flight on the fifteenth of July from New Mexico to the UK.
Telemetry.
People track the flights. That's how open source works now. And then in October, the Federation of American Scientists put out a report arguing the weapons may not actually be present yet, because the base might need security upgrades until twenty thirty-one.
So the experts disagree about whether the bombs are there.
That's a live disagreement. One camp says the flight pattern is conclusive. The other says the infrastructure isn't ready.
Which tells you something about how much of this is inference.
Almost all of it. Nobody announces this.
So walk me through the inference. How do you get from a C-17 flight to "there are nuclear weapons at Lakenheath"?
You look at the aircraft, the route, the timing, the historical pattern of similar flights, the base's known storage infrastructure, and the political context. No single piece is conclusive. The combination is persuasive to some analysts and not to others.
And that's the same method that produces the warhead counts.
It's the same method all the way down. You're always reasoning from traces.
Why does the US do it at all?
NATO's own framing is that nuclear sharing ensures the benefits, responsibilities, and risks of nuclear deterrence are shared across the alliance. That's the official line. A scholar at Leiden, Linde Desmaele, argues the weapons are primarily tools of alliance management, serving three purposes. Deflection, which is a low-cost way for allies to maintain alliance standing. Legitimation, which unifies allies against joining the Treaty on the Prohibition of Nuclear Weapons. And consultation, which gives non-nuclear allies insight and influence over US policy and anchors the US to Europe.
So the bombs are the excuse for the conversation.
The bombs are the price of admission to the conversation. That's the argument.
Now the mechanism, because this is where Daniel's trust question actually gets answered. The US doesn't hand over the weapons.
The weapons remain under US custody and control in peacetime. They're guarded by US Air Force personnel, Munitions Support Squadrons, stored in WS3 vaults inside hardened aircraft shelters.
WS3.
Weapons Storage and Security System. Vaults built into the floor of the shelter. The aircraft parks over them.
And the locks.
Permissive Action Links. PALs. Electronic locks and codes that prevent unauthorized arming or detonation. The US developed and installed them in Europe starting in the early sixties.
What was the threat model?
Steven Bellovin gave a talk on this at USENIX in two thousand six. The original motivation was preventing unauthorized use by enemy countries, terrorists, rogue or pressured US troops, and our allies.
Our allies is in the list. That's the answer to Daniel's question, isn't it. He says it seems like it would involve a huge amount of trust. It doesn't. It's engineered to not require trust.
That's the crux. This is called non-substantive nuclear sharing, because operational control is never transferred. The host nation can't detonate the weapon unilaterally. They can't arm it. They can't do anything with it without the codes.
And before PALs?
Before PALs, the US used what's called substantive sharing. The dual-key Thor IRBM in the UK under Project Emily. An RAF officer's key began the launch sequence and a USAF officer's key armed the warhead. And there was no real protection from British seizure of the American key.
Two keys, one of which could just be taken.
Eisenhower reportedly wanted to give allies de facto control while retaining titular possession only. Which is a sentence that should keep people up at night.
So the sixties were more dangerous than now.
In this specific respect, yes. The locks got better.
What happens in wartime?
Host-nation Dual-Capable Aircraft would deliver the bombs. F-16s, Tornados, transitioning to F-35As. Use is claimed to require authorization from the NATO Nuclear Planning Group and the US president. Reportedly the UK prime minister as well for Lakenheath.
Claimed.
Claimed. The procedures aren't public. And the NPT compatibility question is worth noting, because the US and USSR negotiated the wording of Articles One and Two in October nineteen sixty-six specifically to make nuclear sharing compatible with the treaty. The claim that these arrangements were secret has been thoroughly debunked.
So the whole thing is legal, public, and locked.
Legal, public, and locked.
Which brings us to Daniel's last question. Among the confirmed nuclear powers, how many choose the distributed architecture?
Two.
Two.
The United States and Russia. Of NATO's three nuclear powers, the US, the UK, and France, only the US provides weapons for sharing.
France has floated it.
In March of last year, Macron said France would consider extending its nuclear deterrent to other European states. But both Macron and his defence minister stressed France retains ultimate control, and explicitly denied this amounted to nuclear sharing. Germany's Friedrich Merz said European weapons could only supplement the US arrangement.
So France wants the conversation without the architecture.
They want the strategic effect without the custody problem. Which is a very French solution.
And the UK?
Historically deployed, stationed, and tested weapons on West German, Singaporean, and Cypriot territory. Does not currently share.
So Russia is the other one. Belarus.
After the February twenty twenty-two referendum removed Belarus's neutrality, Putin announced the first warhead delivery on the sixteenth of June twenty twenty-three. Russia supplied nuclear-capable Iskander-M missiles with warheads of variable yield, five to fifty kilotons, and modified Belarusian Su-25 aircraft for bombs of twenty kilotons or less.
And are the warheads actually there?
As of last year, there's no conclusive open-source evidence warheads are actually stored in Belarus. Lukashenko claimed in December twenty twenty-four that Belarus hosts dozens. Putin emphasized Russian control is maintained. A CIS official described it as a double nuclear button.
A double nuclear button.
That's the phrase.
Which is the same architecture as the US model, just with worse public relations.
Same logic. Different branding.
And the speculative one.
Pakistan and Saudi Arabia. A BBC Newsnight report in twenty thirteen, and Western officials, claimed Saudi Arabia could obtain Pakistani warheads at will. Gary Samore said the Saudis believe they have some understanding with Pakistan that, in extremis, they would have claim to acquire nuclear weapons. CSIS called the BBC report partially incorrect and unverified. And then on the seventeenth of September last year, Pakistan and Saudi Arabia signed a Strategic Mutual Defence Agreement. An attack on one is treated as an attack on both. A senior Saudi official said it encompasses all military means.
Which is not the same as a nuclear umbrella.
It's not the same. It's the kind of language that fuels the speculation without confirming it.
So the answer to Daniel is two, and possibly more soon.
The model is being copied. Russia's Belarus deployment, France's deterrent-extension debate, Poland's aspirations, the Pakistan-Saudi pact. The distributed architecture is spreading even as arms control erodes.
Which is the opposite of what most people assume. Everyone thinks the trend is toward fewer weapons in fewer places.
The warhead count is declining. The number of countries hosting someone else's warheads is not.
There's a detail I keep coming back to. The PAL.
The lock.
The most dangerous object in Europe is secured by an electronic lock and a wire seal, and the original threat model included the host nation. That's the whole trust paradox in one device. You don't trust your ally, so you build a lock. And then you station the weapon on their soil anyway, because the alliance needs it there.
And the lock is the trust.
The lock is the trust. Which is a strange sentence but I think it's true.
Hilbert: The vaults smell like hydraulic fluid and something metallic, and they keep them at eighteen degrees year-round.
You've been in one.
Hilbert: I maintained the environmental control systems. Civilian contractor, US airbase in Europe. I was young, I was an electrical engineer, and my job was the air handling in the WS3 vaults. Nobody ever told me what was in them. Everybody knew.
What was the eighteen degrees for?
Hilbert: That's what the spec said. Temperature and humidity both held tight. The equipment didn't care what it was cooling. It cared that it stayed at eighteen.
Did you ever see the weapons?
Hilbert: Once, during a maintenance drill, I saw a PAL up close. Small black box, keypad, a wire seal on the side. It looked like something you'd buy at RadioShack. That's the thing I remember. Not the bomb. The lock that looked like a calculator.
Bellovin's point exactly. The device is mundane by design.
Hilbert: The guards had a motto. Munitions Support Squadron. We own it, you don't.
That's not a motto, that's a policy statement.
Hilbert: It was on a sign.
So the trust paradox, as we've been calling it, is a sign on a wall and a black box with a keypad.
Hilbert: And a smell. Hydraulic fluid and something metallic. I still notice it in garages sometimes.
The interesting part is that the mundane detail is the security. If the lock looked like a nuclear safety device, it would be a target. If it looks like a calculator, it's just equipment.
Hilbert: Nobody photographs equipment.
That's the whole doctrine in five words.
Which connects to something we said earlier about the numbers. The reason the estimates disagree is the same reason the lock looks like a calculator. The system is designed so that the interesting parts don't announce themselves.
And the eighteen degrees.
The eighteen degrees is the part that stays with me. Somebody wrote a spec, and somebody else maintains it, and that's the actual security posture. Not the doctrine. The spec.
So the cutting-room floor. The thing we didn't get to.
The Ivy Mike test, the first two-stage thermonuclear, was ten point four megatons on the first of November nineteen fifty-two. And Castle Bravo came in at two and a half times its predicted yield because of lithium-7, which nobody had accounted for. The physicists were wrong about their own bomb by a factor of two and a half.
That's the part that should worry people more than the locks.
It's the part that explains why the locks exist.
Here's the forward-looking thought, then. Daniel asked how many nuclear powers choose the distributed architecture, and the answer today is two. But the model is being copied, and the thing that makes it work isn't trust. It's a lock that looks like a calculator and a temperature spec that somebody maintains. If the technology fails, or the political situation changes, or the person whose job it is to hold eighteen degrees stops showing up, the architecture doesn't degrade gracefully. It just becomes what it always was. Nuclear weapons in someone else's country.
The lock is the trust. Until it isn't.
That's the episode. Thanks to Hilbert Flumingtop, our producer, who apparently has been in a vault.
He has been in a vault.
This has been My Weird Prompts. If you found this interesting, rate and review the show. It helps other listeners find us.
We'll be back soon.
See you tomorrow.
