IQM's LUMI-IQ Plan Targets Up to Nine Logical Qubits by 2029

by Arash Shirazi · September 11, 2026

Quantum Computing Correspondent, Frontier Tech Wire

IQM Quantum Computers (Nasdaq: IQMX) set out on September 10 a three-stage upgrade path for LUMI-IQ, the superconducting machine it is to install at CSC - IT Center for Science's data centre in Kajaani, Finland, alongside the LUMI AI Factory. The company's release, carried by HPCwire's off-the-wire desk and by StockTitan under the Espoo dateline, describes a sequence that runs 2027, 2028 and 2029 and finishes with what IQM calls a capability for up to nine logical qubits. Nothing in the announcement is a measurement. Every number in it describes hardware that has not yet been delivered.

Stage one, dated 2027, is an IQM Halocene H4 system of 150 physical qubits, which the company says will carry what it terms early quantum error correction capabilities. The 150 figure is a specification for a machine to be installed, not a count of qubits now running at Kajaani, and the release attaches no gate fidelity, readout fidelity or coherence figure to it.

Stage two, dated 2028, is described as a hardware upgrade aimed at lower logical error rates and at enabling real-time error correction. The Quantum Computing Report's account adds that the 2028 step involves low-latency control. No logical error rate, before or after, is published for this stage in any of the accounts reviewed for this article.

Stage three, dated 2029, is an IQM Halocene H5 installation. Here the company claims fast real-time feedback sufficient to sustain logical qubits, and full logical operations including lattice surgery and T-gate teleportation. The headline capability is up to nine logical qubits encoded in distance-3 surface code and colour code, with the further statement that the platform is capable of encoding a logical qubit at distance 11 in surface code or distance 9 in colour code.

The gap between those two claims is worth spelling out, because they are not the same claim. Tech Times, working through the same announcement, put the arithmetic on the record: a distance-3 surface code, the minimum that can correct any single-qubit error per round, encodes one logical qubit across 17 physical qubits, nine data and eight for syndrome measurement, while distance 11 requires 241 physical qubits per logical qubit. Nine distance-3 patches therefore come to roughly 150 physical qubits, and Tech Times notes that 150 physical qubits can support approximately eight to nine logical qubits at distance 3. A single distance-11 patch costs 241. The two configurations are draws of the same order on one physical qubit budget, and IQM has not disclosed the physical qubit count of the H5 system in any of the material reviewed here - not in its own release as carried by HPCwire and StockTitan, nor in the accounts by The Quantum Insider, Quantum Computing Report or Tech Times. Whether the nine-logical-qubit and the distance-11 configurations are alternatives on the same machine, rather than things it does at once, is not stated.

The status of all of this should be plain: it is a vendor roadmap attached to a public procurement. It is not a preprint, it is not a peer-reviewed result, and it is not an independently verified benchmark. No third party has measured an IQM logical qubit at distance 3 or any other distance and published the outcome as part of this announcement.

The framing also drifted in the space of a day. IQM's own headline calls LUMI-IQ a path to Europe's first superconducting quantum computer with logical qubits. By September 11, Tech Times was describing an error-correction blueprint for Europe's first fault-tolerant quantum system. Distance-3 codes correct a single error. Nine of them, in 2029, is a research instrument for learning how to operate logical qubits inside an HPC centre, which is close to how the Tech Times piece itself characterises it further down, noting that nine logical qubits is short of what commercial applications would need.

On the institutional side, the LUMI AI Factory is co-funded by the EuroHPC Joint Undertaking together with Finland, Czechia, Norway and Poland, and CSC is the host. None of the accounts reviewed disclose a contract value for the LUMI-IQ system, the split between EuroHPC and the participating states for this particular procurement, or the terms of the upgrade options that make the machine upgradeable in the first place.

This article carries no direct quotations from the announcement. IQM chief executive and co-founder Jan Goetz and CSC chief executive Kimmo Koski both gave statements, and the wording is consistent across the versions consulted here rather than in dispute: Goetz framed the value as CSC and the LUMI AI Factory owning an upgradeable quantum computer they can integrate into their own AI and HPC infrastructure, and Koski framed the greatest potential as lying in combining the complementary strengths of AI, high-performance computing and quantum computing. Both are rendered here as indirect speech as a matter of house practice; the exact sentences appear in the StockTitan and HPCwire versions of the release.

IQM itself is a newer public company than the Halocene roadmap suggests. It began trading on the Nasdaq Global Select Market on July 3, 2026 under the ticker IQMX, following a business combination with Real Asset Acquisition Corp., and The Quantum Insider reported a pro forma cash position of 337 million euros at that point and an installed base of 23 quantum computers worldwide, including systems at CINECA in Italy, the Leibniz Supercomputing Centre in Germany and Oak Ridge National Laboratory in the United States.

Set against the corporate roadmap IQM published on November 13, 2024, the LUMI-IQ plan is a narrower object. That roadmap aims at fault tolerance by 2030 with hundreds to thousands of high-precision logical qubits using quantum low-density parity-check codes, logical error rates below ten to the minus seven, and a path to scaling up to one million qubits. LUMI-IQ's 2029 endpoint of up to nine distance-3 logical qubits sits well inside that, but the two should not be read against each other as a shortfall: one is a single customer installation with a fixed delivery schedule, the other is a company-wide technology ambition with no customer attached.

The checkable milestones from here are specific. Whether the 2027 H4 delivery lands at 150 physical qubits and on time; whether the 2028 upgrade arrives with published logical error rates rather than a description of real-time correction; whether IQM discloses H5's physical qubit count; and whether any distance-3 logical qubit at Kajaani is characterised in a paper that someone outside IQM and CSC can check. Until then the honest description of LUMI-IQ is a funded plan with dates on it.