Quantum Internet
A network that distributes quantum states over long distances and links quantum computers: quantum repeaters are still missing.
Open in the interactive tree →Quantum states cannot be copied or amplified, and in optical fiber most photons are lost after a few dozen kilometers. A quantum internet therefore needs quantum memories and quantum repeaters that pass entanglement along step by step. It would network quantum computers and enable eavesdropping-proof communication with mathematical proof.
As of October 2026
In February 2026 Qunnect and Cisco demonstrated entanglement swapping over 17.6 km of deployed fiber in New York City at about 5,400 pairs per hour (polarization fidelity above 99%). Also in February 2026 a USTC team reported memory-to-memory entanglement over 10 km of fiber that outlasts the time needed to create it, a key repeater building block (accepted at Nature). In August 2026 another USTC team published entanglement between two atomic memories through 420 km of fiber, but both memories sit in one laboratory, most of the fiber is spooled, and only about one attempt in a million succeeds. A true repeater delivering usable entanglement between distant sites over hundreds of kilometers has not been shown.
What is missing
- Quantum repeaters with long-lived memories and error correction
- Efficient converters between quantum-memory light and telecom wavelengths
- Losses and rate: about one pair per second at metro scale, far fewer over longer distances
- Standards, protocols and network software for quantum networks
- Applications that clearly beat classical cryptography
Becomes possible once solved
- Distributed quantum computing across several sites
- Provably eavesdropping-proof communication
- Networks of atomic clocks and telescopes with higher precision
Open steps
- Repeater protocols and cutoffs Medium AI leverageChoose when to swap and discard entanglement in repeater chains so that rate and fidelity are best with imperfect memories.
- Memory-to-telecom converters Low AI leverageConvert photons from quantum memories to telecom wavelengths with low added noise and high efficiency so they travel through fiber.
- Long-lived quantum memories Low AI leverageStore entanglement for seconds to hours with efficient read-in and read-out, ideally with error correction.
- Quantum network routing and control Medium AI leverageDesign routing, scheduling and control software for multi-user quantum networks with limited, probabilistic links.
Where AI could help
Low AI leverage. AI can tune repeater protocols and lab setups in simulation, but missing repeaters, long-lived memories and converters are physics and hardware problems.
- Optimize entanglement-distribution policies and memory cutoffs in repeater chains by reinforcement learning
- Search for new optical setups and layouts for generating entanglement
- Tune lasers, cavities and memories automatically in the lab
- Schedule and route entanglement in network software
Shown so far
- In February 2024 (Physical Review Applied) researchers used Q-learning in simulations to find repeater-chain entanglement policies with better waiting time and fidelity than swap-as-soon-as-possible. source
Prerequisites
Unlocks
- Global Quantum Internet2040s?