The term on this page
- PQCpost-quantum cryptography
- algorithms that run on ordinary computers and are believed to resist quantum attack
Read this first
The migration off quantum-vulnerable cryptography is happening now, one server at a time, and almost nobody is keeping score in public. The PQC Observatory does: once a day we open two or three read-only TLS connections to a fixed panel of public hosts, record whether each one negotiates the hybrid group X25519MLKEM768, how its certificate is signed, and whether it offers and honours session resumption, and publish the trend. Read-only, restrained, and open about its method. When this post was written the measurement was a single handshake; resumption was added on 28 August 2026.
You can measure a single site's post-quantum posture in a second: our open-source qProbe does exactly that for endpoints you own. What has been missing is the aggregate, the answer to a simple question a journalist or a board asks: is the web actually adopting post-quantum key exchange, and how fast? The observatory is our attempt to answer it in the open, day over day.
What it measures
- Hybrid key exchange. Whether the server selects X25519MLKEM768, the hybrid that stays secure if either the classical or the post-quantum half holds. This is the single clearest signal that a host has started its migration.
- Certificate posture. The negotiated TLS version and the leaf certificate's signature algorithm and expiry, so the series also tracks the slower-moving question of signature agility.
- Reachability. Whether the host completed a handshake at all, so a drop in coverage never reads as a drop in adoption.
- draft-ietf-tls-ecdhe-mlkem: the specification for the hybrid group the panel measures.
How it works, and what it will not do
The method is deliberately dull, because a measurement you cannot trust is worse than none. Once a day a worker performs one read-only TLS handshake per host, spaced out and rate-limited, with a hard timeout and no retries. It never sends application data, never attempts anything beyond the handshake, and honors a refusal or an opt-out immediately. This is the same class of unauthenticated measurement that public scanners such as SSL Labs and Censys have run for years.
It is a separate, self-contained project, pqc-observatory, rather than a mode of our endpoint prober. qProbe, which sits with the rest of the tooling we publish, is gated behind an ownership attestation because it is meant for endpoints you control; the observatory measures hosts we do not own, so it keeps its own narrower, read-only policy and its own opt-out. The panel and the policy are open, and anyone can propose a host by pull request.
Why keep a public series
Migration timelines are driven by evidence, not exhortation. A single number, published every month, does three things at once: it makes progress legible to people who are not cryptographers, it gives early movers a reason to move before a mandate forces them, and it accumulates into a dataset nobody can backfill later. The trend matters more than any single month, which is precisely why it has to start now. See the harvest-now-decrypt-later argument for why the clock is already running.
The current numbers, the panel, and the full methodology live at quantakrypto.com/observatory. If you run one of the hosts we measure and would like it removed, contact us and we will drop it within one measurement cycle.