Pricing…Open Lab
Method — Algorithm · Protocol · intermediate

Quantum teleportation

Quantum teleportation moves an unknown qubit state from one qubit to another. It uses up one entangled pair of qubits shared in advance, plus two ordinary bits sent over a normal link. The original state is destroyed along the way, nothing travels faster than light, and no matter is "beamed" anywhere. It is plumbing, not magic, and real hardware uses it to move quantum states between chips and across networks.

Speedup: No speedup. It moves 1 qubit state for each Bell pair use…Hardware today: Routine in labs. Satellite links have gone beyond 1,000 km.
Complexity
TaskBest classicalQuantum
Transmit one unknown qubit stateimpossible — measuring destroys it, and no-cloning forbids copying1 Bell pair + 2 classical bits

What problem does it solve?

Say you hold a qubit in some state you don't know. You need that exact state on a different qubit somewhere else, like another chip or another lab. The obvious ideas fail.

  • You can't measure it and send the result. A measurement gives you just one bit and wipes out the amplitudes (the numbers that describe the state).
  • You can't copy it first. A rule called the no-cloning theorem says you can never make a copy of an unknown quantum state.

Teleportation solves this with two things. One is an entangled pair of qubits, shared ahead of time. The other is two normal bits, sent when you are ready.

Why can't a normal computer do this?

A qubit's state is a pair of amplitudes. They can take any value on a smooth scale, like the exact position of a dimmer dial. Writing them down exactly would take endless normal bits. And you can't even learn them from a single copy.

Teleportation gets around the whole problem. The state is moved exactly, and neither sender nor receiver ever learns what it was. The two normal bits that are sent tell you nothing about the state. They come out completely random, whatever state is being sent. That is also why the method can't send a signal faster than light.

How does the protocol work?

It uses three qubits. q0 holds the state to send. (Here we make it with the gate RY(0.9).) q1 is the sender's half of a Bell pair. q2 is the receiver's half.

  1. Share entanglement. Put H on q1, then a CX from q1 to q2. That makes a Bell pair. This can happen long before there is anything to send.
  2. Bell measurement. The sender puts a CX from q0 to q1, then H on q0, and measures both. Two normal bits come out. Each of the four possible results is equally likely: 1 in 4, or 25%.
  3. Fix up. The receiver applies X to q2 if the q1 bit was 1. Then they apply Z if the q0 bit was 1. After that, q2 holds the original state exactly. q0 no longer does.

Until the two bits arrive, the receiver's qubit looks like pure noise. Think of a locked box sent ahead of time, and a key sent later by mail. Without the key, the box is useless. Unlike a box, though, nothing solid moves when you use the key. The two normal bits are not just paperwork. They are the step that makes the state usable, and the reason cause still comes before effect.

One honest warning about the demo below. Our circuit format can't run a gate based on a measured bit (this is called classical control). So the fix-up gates can't be applied only when needed. The demo stops just before them. You see the state before the fix-up, grouped by measurement result.

All four (q0, q1) results are equally likely. In shots where q1 = 0, q2 already matches the sent state (about 19% ones). Where q1 = 1, it arrives bit-flipped, waiting for the X fix-up.standby
12345q0|0⟩q1|0⟩q2|0⟩RYHH
press run to acquire
|000⟩|001⟩|010⟩|011⟩|100⟩|101⟩|110⟩|111⟩
————————
counts: sampledamplitudes: statevector, exactengine: in-browser
Open in the Lab →

What is it not?

It is worth saying plainly, because this method attracts more myths than any other.

  • Not faster-than-light messaging. Nothing useful arrives until two normal bits travel by normal means.
  • Not copying. The sender's qubit ends up as a measurement record. Only one copy of the state ever exists.
  • Not moving matter. Only the state, meaning the information, moves. It moves between qubits that already exist at both ends.
  • Not a speedup. It doesn't make any algorithm faster. It matters for how machines are built. Teleportation is how QPU modules get linked, how quantum network repeaters (relay stations) work, and how some gates are run in error-corrected designs.

What does this look like on real hardware?

Teleportation is one of the most tested methods in the field. It was first shown with photons (particles of light) in 1997. In 2017, the Micius satellite mission showed it from the ground to space, over more than 1,000 km. On-chip teleportation that works every time is routine on superconducting and trapped-ion systems. Trapped-ion machines also use a closely related state transfer every day to move information between zones.

Running the full method on a gate-based QPU needs a special feature. The machine must measure in the middle of a circuit and use the result right away to choose the next gate. This is called feed-forward, and several current platforms now support it. Check features on the QPU index. Our in-browser simulator doesn't model feed-forward. That is why the Lab demo honestly shows the state before the fix-up, rather than pretending.

Run the demonstration circuit

Quantum teleportation — demo circuitstandby
12345q0|0⟩q1|0⟩q2|0⟩RYHH
press run to acquire
|000⟩|001⟩|010⟩|011⟩|100⟩|101⟩|110⟩|111⟩
————————
counts: sampledamplitudes: statevector, exactengine: in-browser
Open in the Lab →How would hardware handle it?
Primary sources & further reading