Quantum Key Distribution: Physics as a Tripwire

Quantum key distribution (QKD) lets two people build a shared secret key over a fiber-optic link. It comes with a guarantee. Anyone who measures the light particles (photons) on the way always disturbs them. That shows up as a higher error rate, which gives the spy away. QKD works, and commercial systems exist today at city distances. But it needs special light-based hardware, it is sharply limited by how far light can travel in fiber, and it solves a narrower problem than post-quantum cryptography. That is why most traffic will use PQC instead.

What problem does QKD solve?

Every encrypted conversation starts with a key problem. Two people must end up with the same secret bits, while an attacker watches the channel.

Classical solutions — Diffie-Hellman, and now ML-KEM from the previous chapter — rely on hard math. The spy sees everything but can't compute the secret. Quantum key distribution uses physics instead. It encodes random bits on single photons (particles of light) in a way where measuring them disturbs them. A spy who grabs the photons will always damage a detectable share of them. That is because in quantum physics you can't measure an unknown state without risking changing it. And the no-cloning theorem — a proven result that unknown quantum states can't be copied — stops her from copying photons to study at leisure.

An everyday example: a letter sealed with wax. You can't read it without breaking the seal, and a broken seal shows someone looked. Where the picture breaks: a skilled forger might reseal wax perfectly. With photons, physics itself makes a perfect "reseal" impossible.

Two honest points before the details.

QKD does not encrypt your data. It only makes a shared random key. The data still travels under normal symmetric encryption (usually AES) using that key.
In practice, the safety promise depends on the hardware matching the theory. Real detectors and lasers have flaws that can be exploited. This is an active research area called quantum hacking. The theory's promise is "information-theoretic." That means it holds even against a spy with unlimited computing power, unlike the hard-math promises of classical cryptography. The boxes, though, are engineering.