What exists today of the “quantum internet”?

TL;DR (status as of October 22, 2026): As of October 22, 2026, there is no quantum internet: under the three criteria Rosetta Quantum declares below, zero networks that deliver entanglement between end users are in service anywhere. What is measured is a ledger of dated link demos: entanglement swapping over 17.6 km of deployed New York fiber (Qunnect–Cisco, February 18, 2026), entanglement sharing one fiber with two 800-Gbps classical channels over 24.4 km into Chicago (Optica Quantum, July 2026), a 420 km memory–memory record whose two memories sit in the same laboratory (PRL, August 11, 2026), a five-node swapping relay over 40 km of spooled fiber (arXiv, September 2026) and a trapped-ion repeater node on 17 km of metropolitan fiber (arXiv, October 20, 2026). The largest quantum network actually in operation — China’s 4,600 km backbone with over 150 users (Nature, January 2021) — runs QKD over trusted relays: key distribution, not a quantum internet.

What is the quantum internet supposed to be — and what is it not?

A quantum internet, as the research field defines it, will be a network that distributes entanglement between end nodes, so that any two users can share quantum states on demand — for key exchange with no trusted middleman, for linking quantum processors, for networked sensing. Two boundaries keep the term honest. First, a QKD network is not a quantum internet: QKD distributes classical secret keys, and the deployed backbones do it through trusted relays — nodes where the key exists in classical form and must be guarded by ordinary means — as Rosetta Quantum laid out in its QKD entry. Second, entanglement itself transmits no information — the no-communication theorem, covered here — so a quantum internet will not be a faster internet; it will be a different instrument riding on classical coordination.

Our hero’s zero is a count under three declared criteria: a network (more than one independent user pair), delivering entanglement end to end (not keys through trusted relays), in service (available to users outside the experiment team). Add or remove criteria and the count changes; under these three, as of October 22, 2026, we find zero and we did not find a source claiming otherwise.

What is measured today, link by link?

The table is the answer. Every row is a dated measurement with its source; none of them is a service.

Why is the biggest deployed network not a quantum internet?

Because of what happens at every relay. China’s integrated network — 4,600 km, more than 700 fiber links, two satellite-ground links, over 150 institutional users (Chen et al., Nature, 6-Jan-2021) — is the most impressive quantum-communication deployment on record, and it is a QKD system: at each trusted relay the key is decoded into classical bits and re-encrypted onward. Security inside the relay is institutional, not quantum. That architecture distributes keys at scale — the satellite leg averaged 47.8 kbit/s (same source) — but it cannot deliver entanglement between end users, which is the defining service of a quantum internet. The 2017 Micius satellite demonstrations that preceded it (USTC/Xinhua, 10-Aug-2017) have the same character: historic, measured — and key distribution. This is why our hero plots the 4,600 km bar in gray with an asterisk: measured, deployed, and answering a different question.

What did 2026 actually add?

Four link demos in eight months — 2026 is the densest year on this clock — and each one closes a different gap while declaring its own fine print.

Swapping left the lab (February). Qunnect and Cisco ran entanglement swapping — the primitive that chains short links into long ones — over 17.6 km of commercial fiber between Brooklyn and Manhattan, with room-temperature end nodes, reporting 5,400 swapped pairs per hour over the deployed path against 1.7M+ per hour locally, at >99% polarization fidelity (TQI, 18-Feb-2026; the paper is arXiv:2602.15653). The numbers are company-reported; we print them as such.

Entanglement and your internet shared a fiber (July). The Northwestern–Fermilab team distributed entanglement over 24.4 km from Evanston into Chicago’s StarLight facility while the same fiber carried two 800-Gbps classical channels plus padding power representative of a fully loaded link — quantum photons in the O-band, classical traffic in the C-band — keeping fidelity above 94% (Optica Quantum 4, 342; Northwestern Now, 22-Jul-2026). Teleportation over that field fiber is the declared next step — it will come later; it has not been done there yet.

Memories set a distance record, with an honest asterisk (August). USTC heralded entanglement between two atomic-ensemble quantum memories across 420 km of fiber — the previous matter-to-matter record in fiber was 50 km (Nature 578, 240, 2020, same group) — at a success probability of 1.09×10⁻⁶ per trial and a concurrence of 0.046 ± 0.022, about 2.1 times its uncertainty (PRL 137, 070801, 11-Aug-2026, via PostQuantum). The two memories sit in the same laboratory, with ≈200 km of the fiber per arm on spools and ≈10.1 km per arm deployed: a channel-length record, not a node-separation record — and the paper puts the crossover where memories start beating direct transmission at ≈230 km.

The relay and the node closed the year’s gap from both sides (September, October). Fermilab, Caltech and Northwestern swapped entanglement across a five-node relay — four 10 km segments, spooled — while the same fibers carried 10-Gbps classical data, holding a Bell-state fidelity of 83.7%; their own framing is exact: it is a relay, not a repeater, because it has no quantum memories (arXiv:2609.18899). Two days before this post, the trapped-ion group of B. P. Lanyon and T. E. Northup reported the complementary piece: a true repeater node — two ions in an optical cavity entangling a local photon with a remote one — operating across 17 km of metropolitan fiber at 81.6% Bell-state fidelity (arXiv:2610.27854, 20-Oct-2026, v1, not yet peer-reviewed). Memories without deployment, deployment without memories, and a node without a chain: the repeater that joins all three in the field is the thing that still does not exist in the public record we reviewed.

Who is promising a quantum internet, and for when?

The flagship institutional promise is European: the Quantum Internet Alliance, led by QuTech (TU Delft + TNO) with 40 partners, launched a seven-year program on October 14, 2022 — phase one, 3.5 years and €24M — that will build two metropolitan networks with quantum processors, linked by a long-distance backbone using quantum repeaters (TNO, 14-Oct-2022). The announcement commits to no dated prototype; seven years from launch lands at 2029, and that is our arithmetic, not their promise. On the commercial side, Qunnect calls its February result “a first proof point” toward practical networks (same TQI source) — a vendor saying, correctly, that the service does not exist yet. For what quantum computing itself is used for today, as opposed to networking promises, Rosetta keeps a separate map.

What we don’t know

• Whether the Qunnect–Cisco rates and fidelity hold up to independent review. They come from a company release and an arXiv paper (arXiv:2602.15653); we found no third-party replication.
• The entanglement rates of the Chicago demo. The Northwestern article reports fidelity and the classical load but no pair-generation or detection rates (Optica Quantum 4, 342).
• How robust the 420 km entanglement is. Concurrence 0.046 ± 0.022 is ≈2.1σ, and one of the two heralding channels was statistically indistinguishable from zero (PRL 137, 070801, via PostQuantum).
• Whether the ion-node result survives peer review. arXiv:2610.27854 is a v1 preprint, two days old at publication time.
• Whether a deployed repeater chain exists outside the public record. Our zero counts what is published; a classified or unpublished system would not appear in it.
• Our own “0”. It is Rosetta’s count under three declared criteria (network, end-to-end entanglement, in service), not a community-agreed statistic.

Sources

• The Quantum Insider — Qunnect and Cisco demonstrate metro-scale entanglement swapping over commercial fiber, 18-Feb-2026 (paper: arXiv:2602.15653)
• Northwestern Now — Quantum internet leaves the lab, 22-Jul-2026 (Optica Quantum 4, 342)
• PostQuantum — USTC 420 km quantum-memory entanglement (PRL 137, 070801, 11-Aug-2026; arXiv:2504.05660)
• arXiv:2609.18899 — Entanglement swapping across a five-node relay in a multiplexed quantum-classical network, 16-Sep-2026
• arXiv:2610.27854 — Trapped-ion-based quantum repeater node across a 17 km metropolitan fiber, 20-Oct-2026
• Chen et al. — An integrated space-to-ground quantum communication network over 4,600 kilometres, Nature, 6-Jan-2021 (read via Sci.News, 7-Jan-2021)
• USTC/Xinhua — Micius satellite-to-ground QKD, 10-Aug-2017
• TNO — European program to build world’s first quantum internet ecosystem launched, 14-Oct-2022
• Rosetta Quantum — What is QKD and does your company need it? · What is entanglement and what does it not allow?

Rosetta Quantum publishes verdicts with raw, reproducible data. This is educational content, not a product claim. Quantum, independently measured.