Post-Quantum Has 13 Dated Deadlines. The First Came Due October 22.

Every date checked at its source. The first filings rest on inventories GAO could not fully substantiate. Three more reach suppliers by March.

Qtonic Quantum Research Team | November 7, 2026

We agree with the map. We disagree with its last line, which says the October 22 filings change little outside Washington. They change a great deal, for two reasons the deadline pieces have not connected. The first is what those filings are built on. The second is which of the thirteen dates actually reaches a company that is not a federal agency, and how soon.

The Map

The calendar has thirteen dated entries once you lay the US, NSA and EU instruments side by side [2][3][4][5][6]. Six are binding today. Two are proposed rules with publication deadlines set by executive order, and a third row carries the compliance date that rule would impose. One is NIST IR 8547, which set the industry's planning horizon as a draft from November 2024 and became final on October 27, 2026. Two are agreements between EU governments that are not yet law, and the 2035 line gathers all of the above.

Three things in the map are worth more than the rest. The 2030 and 2031 dates in Section 4(b) of the order apply to high-value assets and high impact systems, not to every agency system, and national security systems follow NSA's calendar rather than OMB's [3][4]. The final NIST IR 8547 is wider than the draft: NIST will deprecate all quantum-vulnerable public-key cryptography after 2030, not only the 112-bit RSA and elliptic-curve parameters, and disallow it under NIST standards and guidelines after 2035. The report also says that window does not relax the earlier dates for high-value assets and high impact systems [5]. And the EU roadmap of June 2025 binds no one, though its first milestone, a defined PQC roadmap in every member state by December 31, 2026, arrives twelve days after the FAR Council's December 19 deadline and ahead of every other American date on the map except October 22 [6].

What the First Deadline Is Built On

Every federal civilian agency was required to file a migration plan with OMB and the Office of the National Cyber Director by October 22 [2]. The plan must include a system prioritization strategy, the methodologies and automated tools used for the cryptographic inventory, and a third-party coordination plan [2].

The memorandum lays out five phases behind that plan. Strategy, planning and discovery in 2026 and 2027. Pilots and early migration in 2027 and 2028. Prioritized migration from 2028 to 2030, when high-value assets and high impact systems move to post-quantum key establishment. Signature migration in 2031. Full migration of remaining systems by 2035, conditioned on risk and on what the market can supply [2]. October 22 is the first deliverable of the first phase, and every later phase is scoped by it.

On October 6, GAO published its audit of how the 24 selected agencies had done on exactly that inventory. One agency fully built it. One had not started. Twenty-two were in between. None of the six agencies GAO sampled could fully substantiate the algorithms their own inventories claimed, and four of the six, in GAO's words, "did not address this activity". Across the 22 inventories, several listed symmetric algorithms as quantum-vulnerable [7]. Those findings date from an audit that closed in September 2025, so an agency may have rebuilt its inventory since. GAO's 89 recommendations are the test of whether it did.

That is the fact the deadline articles leave out. The first deadline on the calendar fell on October 22, and the plans due that day rest on a foundation the government's own auditor had already described as inaccurate or incomplete. A deadline met that way does not settle anything. It starts the next round of questions, and the third-party coordination plan says who will be asked.

The Three Dates That Reach You

Most private companies have no legal post-quantum deadline, and the Technology.org piece is right to say so [1]. Their customers do, and the dates that bind the customer arrive at the supplier as procurement, which is faster than law.

The first is December 19, 2026. Section 6(c) of the order gives the FAR Council 180 days to publish a proposed rule requiring covered contractors to comply with NIST's FIPS, including the post-quantum ones, by December 31, 2030 [3]. The compliance date is in 2030. The rule text, and the questions it puts into federal security questionnaires, is due December 19, 2026.

The second is January 1, 2027. From that date, CNSA 2.0 algorithms are required in all new acquisitions for national security systems [4]. That is ML-KEM-1024, ML-DSA-87 and the rest of the suite, with no credit for larger RSA or elliptic-curve keys, which NSA's own FAQ says do not help [4]. If needed, the FAQ allows a deployment to be accepted on a contractual commitment by a vendor to provide a CNSA 2.0 update, but only "when combined with a realistic deployment plan prioritizing partial QR that meets the 2030 deadline for full QR," and such an acceptance "will need to be regularly tracked and reported" [4]. So the question a prime will put to its subs is not only which algorithms ship today but which are committed under contract, and when [4]. A sub that cannot name its algorithms cannot answer either half.

The third is March 19, 2027. Section 5(d) gives CISA, with NIST, 270 days to publish the minimum elements of a cryptographic bill of materials [3]. Once that exists, asking for one costs a buyer nothing. The same 270 days produce a proposed rule on contractor vulnerability disclosure that must take in reports of cryptographic weaknesses [3].

The dates that say 2030 and 2031 are real, and for the hardware in a program of record they are already inside the refresh cycle. But the dates that will change a supplier's quarter are the three above, and the latest of them falls on March 19, 2027.

For a bank or a hospital with no federal contract, the dates arrive sideways. In our reading, NIST's deprecation schedule, now final, becomes the horizon for every validation program and framework that cites FIPS, and the EU's 2030 milestone for high-risk use cases will be read by national regulators as covering the sectors they already supervise [5][6]. The federal supplier gets a questionnaire in 2027. The regulated enterprise gets an examiner's question a year or two later, and the answer is the same inventory.

Why 2030 Is a Decision This Year

The usual objection is that no cryptographically relevant quantum computer exists, which is true. The Global Risk Institute's 2025 survey of 26 experts puts the chance of one within ten years at 28 to 49 percent, and within fifteen years at 51 to 70 percent [8]. In March, Google Quantum AI published estimates that breaking a 256-bit elliptic-curve key would take fewer than 1,200 logical qubits and fewer than 500,000 physical qubits, running for minutes [9]. Google has set 2029 for its own migration [10]. Cloudflare set the same year for itself. On April 7, 2026, it said it now targets 2029 "to be fully post-quantum (PQ) secure including, crucially, post-quantum authentication" [11].

Those are forecasts and engineering targets, not events, and we treat them as such. The arithmetic that turns them into a decision is older than any of them. Add how long the data must stay secret to how long the migration will take. If the sum crosses the earliest credible arrival date, the system is already exposed, because an adversary who records the traffic now only needs the machine to arrive before the secret expires.

Five years is an optimistic migration estimate. GAO's audit found the largest agencies nearly three years into the inventory step with 22 of 24 still incomplete [7]. On that arithmetic, patient records, firmware signing keys and defense design data are all past the line today, and no 2030 deadline moves them back.

What We Are Not Saying

We are not predicting a CRQC date. The survey ranges are the experts' and the 2029 figure is Google's engineering target for its own systems, not a forecast [8][10].

The EU dates are not law. The roadmap is coordination among member states. A proposal to write a post-quantum policy requirement into the NIS2 directive exists, and it has not passed [1][6].

IR 8547 is NIST guidance, not law. It became final on October 27, 2026. It sets what NIST standards and guidelines will accept, and it does not by itself bind a private company [5].

We cite Google's and Cloudflare's 2029 dates as the migration targets those companies set for their own systems. We do not present either as a date for a working attack.

Where to Start

Two of the thirteen dates ask for things that exist today.

The first is the inventory, which every other date depends on. A QScout assessment produces the artifact the October 22 plans were supposed to rest on: a prioritized inventory of systems with vulnerable cryptography, across a governed 74-module catalog, delivered where approved as a cryptographic bill of materials in CycloneDX 1.7 [12]. When the CBOM minimum elements are published, due March 19, 2027, that is a format a questionnaire can ask for by name.

The second is the data that is already past the line in the arithmetic above, on systems that cannot be re-engineered before 2030. QShield by Qtonic Quantum Corp. is software only, with no hardware changes: post-quantum protection for the connections between enrolled Linux hosts, meant to address harvest-now-decrypt-later exposure for traffic routed through its protected path. It does not recover data recorded before protection began [13]. The cryptographic profile is locked at ML-KEM-1024 key establishment, ML-DSA-87 signatures and AES-256-GCM with no classical fallback, which are among the CNSA 2.0 algorithms [13]. The evidence is company-published, not independently verified, and hashed: a completed 72-hour run between two hosts with zero failures [13]. Separately, a signed historical record lists all 50 locations from September 26 to 28 [13]. The qualifying long-duration soak has not started, and its status is published with the evidence [13]. It is delivered through paid, operator-led engagements on Linux hosts the customer already runs [13].

Both are governed engagements rather than downloads. An analyst confirms the requester and the scope before any work begins [12][13].

This material is for informational purposes only and does not constitute legal, regulatory, compliance, or professional advice, and it does not establish any professional relationship. Dates and requirements are summarized from the public instruments cited and may be amended. Day counts from signing dates to rule deadlines are our arithmetic. QShield by Qtonic Quantum Corp. is not represented as FIPS 140-3 validated or as holding Government authorization. Results described are those of the specific runs and records cited, within their stated scopes and dates. References to Technology.org, OMB, ONCD, NSA, NIST, CISA, GAO, the European Commission, Google, the Global Risk Institute and other third parties are based on publicly available information and are used for analysis and commentary. They do not imply partnership, endorsement, or affiliation. Product capabilities are described as of the date of publication and apply within the authorized scope of an engagement.