Quantum key distribution (QKD) patents: open strategic opportunities
The USPTO has granted more than 600 QKD patents and the EPO about 180. What those patents cover, who holds them, and where the field is still open.
The open space in quantum key distribution (QKD) patents is narrowing. QKD filings at the United States Patent and Trademark Office (USPTO) more than tripled between 2016 and 2024. The USPTO has granted more than 600 QKD patents, and the European Patent Office (EPO) has granted or approved about 180. Every one of those grants, and every published application behind them, is prior art against the next filing.
The patent record shows which parts of the field are already claimed and which are still open. For a QKD company, that is a map of where a new filing can still build a position.
High grant rates, but not easy patents
The USPTO has granted about 90% of QKD applications that have reached an outcome: 603 granted and 69 abandoned. At the EPO the figure is 83%: 181 granted or approved for grant, 37 withdrawn and one refused.[1]
A high grant rate is not the same as broad protection. The rate counts every application that ended in a grant, however narrow the claims that survived. Huawei’s US patent, for example, was granted only after its claim was amended to keep the quantum signal “only in an S band” (see below).
Filing has grown at both offices. USPTO QKD filings rose from 28 in 2016 to 91 in 2024, after a peak of 113 in 2023. EPO filings rose from 13 in 2016 to 53 in 2023. Each of those filings, once published, adds to the prior art the next applicant has to get past.
Few appellate decisions, and none on protocols
Applicants who took a rejection to appeal have done worse. The USPTO’s Patent Trial and Appeal Board has decided six appeals on QKD-type inventions, and the applicant lost all six on prior art: earlier publications already disclosed what was claimed, or made it obvious.
Five of those appeals, decided between 2009 and 2015, concerned optical details. An optical layout, a timing choice, an alignment correction and a noise filter were each found in earlier optical systems.
The sixth, decided on August 21, 2026, came closest to a protocol. Leidos claimed a system that interleaves single-photon qubits into a classical optical message in a predetermined pattern, so that tampering shows up in the qubits. The Board held the claim anticipated by an earlier publication that multiplexed single photons into a classical signal according to a pre-negotiated pattern to detect eavesdropping. Even there, the question decided was novelty, not whether such a step can be patented.
The EPO’s Boards of Appeal have issued two appellate decisions near the field. An entangled-photon source from France’s CNRS was granted only on the cavity structure the applicant had disclosed, and a Novomatic quantum random number generator was treated as ordinary photonics.
Neither office’s appeal body has decided whether a QKD protocol step can be patented. That question is still open.
What the claims cover
The two offices grant different kinds of QKD patent. The table compares the patents each office has granted.
| Main subject of the claims | USPTO QKD grants | EPO QKD grants |
|---|---|---|
| Protocol steps (sifting, reconciliation, privacy amplification, decoy states) | 31% | 34% |
| Optical hardware (sources, detectors, lasers, interferometers) | 24% | 38% |
| Channel and transport (fibre, co-propagation, satellite) | 20% | 25% |
| Network and key management (relays, trusted nodes, key managers) | 17% | 14% |
| Application layer (payments, authentication, blockchain, 5G) | 20% | 7% |
A claim can fall under more than one heading, so the columns add to more than 100%.
Across all USPTO QKD applications, granted or not, 57% recite neither a protocol step nor optical hardware. Of those, 44% claim the network around QKD, the management of the keys it produces, or what is done with them; 11% claim the optical channel; the rest describe a QKD system in general terms.
EPO grants lean the other way, toward the physics. Hardware and protocol together dominate, and application-layer claims are rare. Part of the difference is who files where: the US banks that drive application-layer filing at the USPTO file almost nothing at the EPO.
Who holds them
At the USPTO
The largest QKD filers are Toshiba (73 applications), universities (56 together), Wells Fargo (25), NEC (22), and Bank of America and Alibaba (21 each).
Banks and payment companies are a distinct group. Wells Fargo, Bank of America, JPMorgan and Mastercard together hold 57 QKD applications, and 24 of them claim the application layer only. Their filing has picked up: 27 of the 57 were filed in 2023 and 2024 alone. A bank or payment company planning a QKD deployment should check those claims first.
At the EPO
The list is equipment makers and telecom operators: Toshiba (37), ID Quantique (29), Huawei (28), universities (22), BT (16), Deutsche Telekom (15), Terra Quantum (11) and Arqit (10).
The US banks have filed almost nothing in the EPO’s QKD and quantum transmission groups: two Mastercard applications, and none from Wells Fargo, Bank of America or JPMorgan. A company designing a new transmitter, detector or reconciliation step for Europe is working where Toshiba, ID Quantique and Huawei hold the most applications.
The same invention at both offices
Four quantum-communication applications refused by the EPO highlight the gap between the offices. Only one, QuantumCTek’s, is a QKD application by the test used in this article. The others concern the fibre that carries the quantum signal, a quantum random number generator and the relaying of QKD keys.
- QuantumCTek’s QKD terminal was refused at the EPO as no more than “an alternative way to integrate the components.” The USPTO allowed it after a single office action that raised only two wording defects, with no prior art applied.
- Huawei’s scheme for running quantum and classical signals in the same fibre was refused at the EPO. The USPTO granted it after the claim was amended to keep the quantum signal “only in an S band.”
- The Université libre de Bruxelles’s networked random number generator was refused at the EPO and granted at the USPTO with its independent claims unamended.
- Deutsche Telekom’s key relay was refused at the EPO; we have not identified a US counterpart.
For a company filing at both offices, these cases argue for drafting the specification with the EPO’s reading in mind from the first filing.
Where the strategic openings are
Hardware: valued, and being bought
The EPO’s grants show where European protection is most readily available: hardware, and protocol steps tied to the optics. Larger companies have started buying the hardware makers that hold those patents. IonQ bought Qubitekk and ID Quantique, and listed ID Quantique’s nearly 300 granted and pending patents among what it acquired. China Telecom agreed in 2024 to take a majority stake in QuantumCTek.
The protocol layer: open, for now
A company that wants a position beyond the box has the protocol layer, where neither office’s appeal body has yet ruled on what can be patented. That opening will not stay open indefinitely. Each year’s filings add to the prior art a protocol claim has to clear.
Application-layer rights for banks and telecom operators
For a bank or telecom deploying QKD, the US record matters in a different way. Banks are filing at the USPTO on how QKD keys are used in payments and authentication. Knowing who holds those application-layer claims is part of planning a deployment.
Drafting implications
Two lessons from the appellate decisions apply. First, the specification should define every quantum term. In a USPTO appeal on a quantum optical interconnect, an undefined “quantum state” was read onto classical polarization, and the claim fell to classical art.
Second, claims should identify the structure that produces the quantum property rather than the property alone. That is the line the EPO Board of Appeal drew in the CNRS appellate decision.
Note on method
[1] How we identified QKD patents. The starting point was the patent classification group for quantum cryptography: 2,274 applications at the USPTO, and 891 in the EPO’s QKD and quantum optical transmission groups. Classification alone overstates the field. At the USPTO only about 1,080 of the 2,274 applications (47%, plausible range roughly 900 to 1,200) are about QKD; the rest are quantum networking and quantum memories, post-quantum (classical) cryptography, quantum random number generators, ordinary key management and optical communication, and a few unrelated filings. At the EPO about 350 of the 891 meet the same test on their titles and claims; we did not read samples there, so that figure is a floor.
We sorted each application by its title and independent claims. An application counted as QKD if it referred to quantum key distribution, a quantum key, quantum cryptography, or a named QKD protocol or technique, or if it combined a quantum signal (photons, entanglement, a quantum channel) with a cryptographic key. We then examined samples from this pool, in three groups. In the first, the claims referred to QKD but not to post-quantum cryptography, and every sampled application was in fact about QKD. In the second, the claims referred to both QKD and post-quantum cryptography; about a third of the sampled applications were about QKD, and the rest were about post-quantum cryptography. In the third, the claims used none of the QKD terms; about one in five sampled applications were still about QKD, mostly QKD hardware whose claims never mention a key. The total is an estimate built from those samples, and the range reflects their size.
The grant rates and filing counts in this article use the applications that met the test from their claims or titles and had been published by August 30, 2026: 805 at the USPTO and 349 at the EPO. The holder counts and the claims breakdown use the USPTO applications that met the test from their claims (651). USPTO outcomes are counted as of August 30, 2026. The EPO Register data we used records current status without dates, so EPO outcomes may include a small number of later grants. The breakdown of what the claims cover is based on the same keyword reading of the independent claims, so a claim can count under more than one heading.
Source: CipherLaw; based on the USPTO’s records for 2,274 applications in the quantum cryptography classification group, the EPO’s records for 891 applications in the QKD and quantum transmission groups, and the appellate decisions at both offices.