Europe now has a row of quantum computers with names. PIAST-Q in Poznań, VLQ in Ostrava, Jade near Cologne, Ruby outside Paris, Euro-Q-Exa in Munich, Lucy at the CEA's computing centre south of Paris. Between June 2025 and April 2026, each was inaugurated with ribbons, speeches and a press release; the last, Lucy, is a photonic machine planned with 12 physical qubits. A reader who counts those ceremonies will conclude that Europe is delivering quantum machines on schedule. A reader who reads what each document actually says will find something more precise and more useful: the dates are real, but almost all of them date a ceremony, and none of them dates the day a researcher was let in to run a job.
This piece belongs to the quantum chapter of the CareerOn Industry Atlas. The first piece followed the money and found the French State acting as planner and buyer. The second followed the hardware underneath and found three layers France does not yet fully supply. This one asks what is actually being built, and when. It reads only documents written by the programmes that buy the machines and the centres that host them, never a vendor announcement on its own, and it keeps four words apart that the industry tends to blur: delivered, inaugurated, operational and open to users.
Europe's quantum build is real and dated, but what it dates is mostly ceremony: five public machines carry a dated inauguration and only one carries a dated physical delivery, no document dates the day researchers were admitted, no slip can be measured because every later document records a different stage from the plan before it, and France is funding the means to make quantum chips as well as the machines; for a candidate, the useful skill is telling those stages apart, and the work sits in the months between a ceremony and a user.
We read ten dated documents from three kinds of publisher. Six come from the European programme that buys the machines, the EuroHPC Joint Undertaking, and from the HPCQS project it funds: the project's launch release of December 2021, its joint statement with the hosts and the vendor of November 2023, the Lucy contract notice of 26 September 2024, the inauguration notices of 13 November 2025, 12 February 2026 and 14 April 2026, and the Joint Undertaking's annual accounts for 2024, published in June 2025. One comes from a host, the CEA, dated 23 October 2025. One is the national audit, the Cour des comptes report of July 2026, read again for a passage the earlier pieces did not use. One comes from the FAMES pilot line in Grenoble, dated 28 September 2026. Four documents were refused, among them the Joint Undertaking's own catalogue of its machines, because it is a living page with no date, and we could not say when any line on it was true.
What has a date
Start with what can be counted. The Joint Undertaking signed its contracts first. Its accounts for 2024 say that "all procurement procedures were finalised in 2024", that four contracts were signed that year, including the French one, and that the Spanish and Italian contracts followed in early 2025. A signed contract is a commitment to deliver. It is not a machine, and we never add it to one.
Then come the ceremonies. The notice of February 2026 says PIAST-Q was inaugurated in Poznań in June 2025 and VLQ in Ostrava in October 2025, before Euro-Q-Exa in Munich that day. Jade and Ruby, the two neutral-atom machines of the HPCQS project, were inaugurated together in November 2025. Lucy, the photonic machine at the CEA's computing centre, was inaugurated on 14 April 2026, and the notice says plainly that "Lucy is the fourth EuroHPC quantum computer to be inaugurated". That is five named machines, or six counting Jade and Ruby separately, each with a dated inauguration printed by the programme that paid for it.
Five machines carry a dated ceremony; one carries a dated delivery
| Machine, host country | State the document records | When | Publisher |
|---|---|---|---|
| PIAST-Q, Poland | Inaugurated | June 2025 | EuroHPC |
| VLQ, Czechia | Inaugurated | October 2025 | EuroHPC |
| Jade and Ruby, Germany and France | Inaugurated | November 2025 | EuroHPC |
| Euro-Q-Exa, Germany | Inaugurated | February 2026 | EuroHPC |
| Lucy, France | Delivered, in acceptance | October 2025 | CEA |
| Lucy, France | Inaugurated | April 2026 | EuroHPC |
Sources 6 European High Performance Computing Joint Undertaking · 7 European High Performance Computing Joint Undertaking · 5 European High Performance Computing Joint Undertaking · 4 Commissariat à l'énergie atomique et aux énergies alternatives
Only one document dates a physical delivery. On 23 October 2025 the CEA wrote that Quandela had delivered Lucy to its computing centre, and that "the system has entered an acceptance phase before its opening to European researchers at the beginning of 2026". Acceptance is the stage where the buyer tests that the machine does what the contract says. It is the most informative sentence in the whole set, because it names a state between delivery and use, and it is the only one that does.
What no document dates is access. The notice of February 2026 adds that "four are operational, and two more are under procurement", but it does not say which four, and it does not say when a researcher outside the host could first submit a job. The hypothesis we started with, that at least three European machines carry a delivery date in an official document, holds if a delivery means an inauguration. Read strictly, it holds once. We print both readings, and we say plainly that a ceremony is not access.
The counter-case is fair. Programmes inaugurate machines when they work, and a host does not invite a minister to switch on a box that does nothing. An inauguration is evidence that something runs. It is not evidence of what runs, for whom, or how often, and a candidate deciding whether to join a host centre needs the second kind of evidence, not the first.
The plan beside the record
The second hypothesis was that at least one delivery date has moved. We set a strict rule before reading: a slip counts only when two dated documents give the same delivery, in the same state, two different dates. The documents do not meet it.
Take Jade and Ruby. In November 2023 the hosts and the vendor wrote, together, that "the two PASQAL quantum computers will be accessible to a wide range of European users in 2024". The next dated document, two years later, records an inauguration. Accessible to users and inaugurated are different states. No document we admitted says whether users were admitted in 2024, so we cannot say the plan was missed, and we cannot say it was met.
Take Lucy, which has the longest paper trail. In September 2024 the programme wrote that "the installation of the system will start in 2025". In October 2025 the host wrote that the machine had been delivered and was in acceptance, with an opening to researchers planned at the beginning of 2026. In April 2026 the programme recorded an inauguration. Each document is consistent with the one before it, and each names a different stage. A reader in a hurry will call Lucy late, since April is not the beginning of the year. The documents do not say that. They say an inauguration happened in April; they do not say when researchers came in.
What was planned, and what was later recorded, are different states
| Machine | Planned | Planned in | Later recorded | Recorded in |
|---|---|---|---|---|
| Jade and Ruby | Accessible to European users in 2024 | 2023, hosts and vendor | Inauguration | 2025, EuroHPC |
| Lucy | Installation to start in 2025 | 2024, EuroHPC | Delivered, acceptance phase | 2025, CEA |
| Lucy | Opening to researchers at the beginning of 2026 | 2025, CEA | Inauguration | 2026, EuroHPC |
Sources 2 Commissariat à l'énergie atomique et aux énergies alternatives · 5 European High Performance Computing Joint Undertaking · 3 European High Performance Computing Joint Undertaking · 4 Commissariat à l'énergie atomique et aux énergies alternatives · 7 European High Performance Computing Joint Undertaking
So the hypothesis fails, and this piece refuses to call the gap a delay. That is not generosity to the programme. It is the same rule we would apply to a company: a gap between a plan and a later record of a different stage is a gap in the record, and the honest response is to ask for the missing document, not to estimate the missing date. The counter-case is that this rule lets a programme hide behind its own vocabulary. It can, and that is why the next thing a reader should ask of each host is one dated sentence: the day the first outside user ran a job.
Building the means, not only the machines
The third hypothesis was that France is building capacity to make quantum hardware, not only buying finished machines. Two documents support it, and they date it differently.
The Cour des comptes, in July 2026, says that 92 % of one national project, CEPREQ, went to setting up a factory for Pasqal, and adds that two French companies have already delivered and commissioned several machines at host sites, including the CEA's computing centre. The audit prints a spending share. It prints no opening date for the factory, so it proves that France paid for production capacity, not when that capacity produces anything.
The FAMES pilot line in Grenoble, run by CEA-Leti, dates its own step. On 28 September 2026 it announced that it gives Quobly access to process capabilities to support "future generations of silicon quantum processors". That is a capability, dated: a line where silicon qubits can be made with the tools of the chip industry. The same release carries a company date, that "Quobly plans to commercialize its first quantum computer, Alloy Pioneer, by the end of 2026". We print it as what it is, a company claim carried in a partner's release, and not as a delivery by any programme.
France funds the means to make quantum chips, not only the machines
| What is built | What the document prints | Kind of claim | Publisher, date |
|---|---|---|---|
| A Pasqal factory | 92 % of the CEPREQ project spent on setting it up; no opening date | Audit finding | Cour des comptes, July 2026 |
| Access to the FAMES pilot line | Process capabilities for future silicon quantum processors | Programme agreement | FAMES Pilot Line, September 2026 |
| A first Quobly computer | To be sold by the end of 2026 | Company claim | FAMES Pilot Line, September 2026 |
Sources 9 Cour des comptes · 10 CEA-Leti
The counter-case is that a factory and a pilot line are inputs, and inputs are easy to fund and hard to turn into machines. That is true, and the first piece of this chapter showed how much of the French effort is still a plan. But the distinction matters for this piece, because a country that only buys machines builds operators, while a country that also builds lines builds engineers. France is doing both, and only the second is dated in the record we read.
What this means for your career
The first lesson is vocabulary, and it is worth more than it sounds. Delivered, in acceptance, inaugurated, operational and open to users are five different stages, and each has its own people. Acceptance belongs to the engineers who test a machine against its contract. Operational belongs to the host's operations team. Open to users belongs to the people who write the access rules, the software stack and the support desk. A candidate who can say which stage a machine is at, and which document dates it, sounds like someone who has worked inside a host centre.
The second lesson is where the work sits. The months between a ceremony and a user are not empty. They are integration with the supercomputer next door, calibration, scheduling, and the first hybrid jobs. That is where host centres hire, and it is the stage no press release describes. The third lesson is geographic: the Grenoble line puts silicon quantum chips in the same place as the French chip industry, so a process or device engineer in that region is closer to quantum work than their title suggests.
The simulation Cryogenic electronics engineer: integrated quantum control at 4 K is placed in this piece, against the pilot line's work on silicon quantum processors, which is the route this simulation rehearses on the cold side. One honest caveat: the release names process capabilities, not control electronics, so the simulation practises the cold side of that route, not the fabrication itself. The simulation Superconducting qubit engineer: designing and characterising a cat qubit was placed in the first two pieces; none of the ten documents in this one names a superconducting machine or line, so we print it as unplaced here. We have no simulation yet for acceptance testing, for host operations or for hybrid scheduling, and all three are recorded as gaps in our catalogue.
One question to ask a host centre or a vendor, made possible by this piece: on which day did the first user outside your own team run a job on this machine, and which document says so? A clear answer is rare, and a company that can give one is a company that measures what matters.
What we could not prove
We found no admitted document that dates user access to any of the six machines. We could not tell which four machines the programme counts as operational. We could not date the opening of the Pasqal factory, and we could not read the Chips Joint Undertaking notice on the pilot lines in a form we could check, so it is not used. The project launch of December 2021 gives a duration of 48 months, but it prints no date for either machine, so we do not measure anything against it. If a host publishes a dated access record, we will add it and say whether it turns any of these gaps into a measured delay.
The next piece of the chapter turns from what is built to what is argued: which kind of machine is ahead, when a useful machine arrives, and whether error correction is still science or now engineering.