Iqm.tech

By the end of 2029, Finland's Kajaani data center will house something that does not yet exist anywhere in Europe: a publicly owned, nationally funded superconducting quantum computer explicitly designed to sustain error-corrected logical qubits — the form of quantum information stable enough to run real fault-tolerant algorithms. IQM Quantum Computers (Nasdaq: IQMX) announced the precise three-phase upgrade architecture for LUMI-IQ on September 10, 2026 — naming the error correction codes, the logical qubit capacity, the gate operations, and the hardware milestones that define Europe's formal commitment to the fault-tolerant computing era. The announcement covers the quantum system selected by CSC – IT Center for Science for the LUMI AI Factory complex in Kajaani, and was published in full as an IQM press release via BusinessWire.

From Physical Qubits to Logical Qubits: Why the Distinction Defines the Field

To understand what LUMI-IQ represents, a reader must first understand the distinction the announcement is built around. Every quantum computer built today — including IQM's own 20-qubit Pathfinder system deployed at Oak Ridge National Laboratory in June 2026 — operates on physical qubits: individual superconducting circuits cooled to roughly 10 millikelvin (−459.6°F), colder than outer space, that lose their quantum state to environmental noise within microseconds.

A physical qubit is fragile. Any calculation that requires more quantum operations than a single qubit can sustain before noise corrupts its state will fail. This is the regime nearly every deployed quantum computer lives in today — the Noisy Intermediate-Scale Quantum (NISQ) era.

A logical qubit is a different thing entirely. It encodes one reliable unit of quantum information across many physical qubits working in concert, using quantum error correction to continuously detect and fix errors without destroying the encoded state. The most widely deployed approach is the surface code: physical qubits arranged in a two-dimensional grid, with continuous "syndrome measurements" that catch errors in neighboring qubits before they propagate. For a distance-3 surface code — the minimum that can correct any single-qubit error per round — that means encoding one logical qubit across 17 physical qubits (9 data + 8 ancilla for syndrome measurement), a technique introduced through lattice surgery and now foundational to the field.

Distance is the key parameter. A distance-3 code corrects 1 error; a distance-5 code corrects 2; a distance-11 code can correct 5. Error suppression improves exponentially as distance increases — but so does the physical qubit overhead per logical qubit. Distance-11 requires 241 physical qubits per logical qubit. That overhead is why building even a few reliable logical qubits remains one of the hardest engineering problems in science.

The transition from NISQ to logical qubits is, broadly, the transition from quantum experiments to quantum utility. LUMI-IQ is Europe's first publicly funded, nationally owned commitment to making that crossing in superconducting hardware, as confirmed on the LUMI project's official announcement page.

Three Phases, One Ownership Model

LUMI-IQ will be delivered through IQM's Halocene product roadmap in three distinct phases, each building on the previous without requiring a system replacement. The Halocene product line, announced in November 2025, is an open, modular platform for quantum error correction research.

Phase 1 (2027): The IQM Halocene H4 system arrives at CSC's new data center in the Renforsin Ranta business park in Kajaani — a facility built on the site of a former paper mill and currently under construction — with 150 physical qubits and early quantum error correction code testing and calibration infrastructure.

Phase 2 (2028): A hardware and control electronics upgrade lowers logical error rates and introduces real-time error correction feedback loops. This step is critical: error correction without fast classical feedback cannot sustain a logical qubit across a full computation. The upgrade brings LUMI-IQ into the regime where logical qubit preservation becomes practical rather than experimental.

Phase 3 (2029): Installation of the IQM Halocene H5, IQM's early fault-tolerant quantum computing platform. Upon completion, LUMI-IQ will support computations with up to 9 fault-tolerant logical qubits using distance-3 surface code and color code implementations. The architecture is also designed with headroom to encode single logical qubits up to distance-11 surface codes and distance-9 color codes — a structural capability that far exceeds the 9-qubit operational limit but provides researchers with a platform to study higher-distance encoding in controlled conditions.

Alongside those 9 logical qubits, Phase 3 will execute active lattice surgery — the practical method for performing logical two-qubit gate operations by temporarily merging adjacent qubit patches and measuring joint stabilizers — and T-gate state distillation. That second operation matters especially: the T gate is required for universal quantum computation but cannot be implemented directly on surface codes. Instead, it must be synthesized through distillation of "magic states" — a resource-intensive procedure that demands additional physical qubits dedicated to generating and purifying ancilla quantum states. The inclusion of T-gate distillation in LUMI-IQ's design, first formally described by Bravyi and Kitaev in 2005, signals that the Halocene H5 is designed for universal fault-tolerant quantum algorithms, not simply error-correction demonstrations.

The co-funding structure ties four nations to this roadmap. LUMI-IQ is jointly funded by the EuroHPC Joint Undertaking and a consortium of Finland, Czechia, Norway, and Poland. Critically, the system will be owned outright by the participating nations and hosted on Finnish soil. This on-premises ownership model is central to IQM's commercial strategy — the national consortium controls research scheduling, retains intellectual property generated on the hardware, and owns the upgrade pathway. IQM states it has sold 23 quantum computers globally — more than any other manufacturer, it claims — with LUMI-IQ representing the company's most technically ambitious deployment to date.

The Research That Makes the Roadmap Credible

What distinguishes IQM's Halocene approach from aspirational roadmaps is a body of hardware-grounded research that predates the LUMI-IQ commitment. In June 2026, IQM researchers — working with teams at Freie Universität Berlin, the University of Edinburgh, and Johannes Gutenberg-Universität Mainz — published a preprint introducing "directional tile codes": a new family of quantum low-density parity-check codes specifically designed to run on IQM's existing square-lattice hardware architecture.

The result was striking. By exploiting IQM's native iSWAP gate connectivity and using dynamic syndrome extraction circuits, the directional tile codes reduced per-logical-per-round error rates by up to 1,000 times compared to standard surface codes — at a hardware footprint of roughly 30 physical qubits per logical qubit, compared to 17 for a distance-3 surface code but with substantially higher logical fidelity. The codes require no new hardware or connectivity changes; they run on the same Crystal processor architecture IQM already manufactures and ships.

Halocene H4 and H5 will deploy surface and color codes, not directional tile codes — the more conservative, well-validated choices for a first-generation production system. But the directional tile code research demonstrates that IQM's hardware is compatible with the next generation of QEC approaches, and that the company is not simply licensing an established error-correction protocol. It is developing error correction co-designed with its own hardware. That distinction matters for a long-duration upgrade program: researchers at LUMI-IQ can expect to run experiments on the hardware that generate foundational results, not merely replicate what better-resourced programs have already demonstrated elsewhere.

"At IQM, we have always believed that building production-grade quantum systems and advancing the underlying science are two sides of the same mission," IQM stated in the directional tile codes BusinessWire announcement. "Close collaboration with leading academic groups is central to that approach."

What Nine Logical Qubits Can and Cannot Do

The most important calibration this article can offer is also the one most articles about quantum milestones omit: 9 logical qubits is not enough to solve commercially valuable problems.

Current estimates for the logical qubit counts required for the most consequential quantum applications are significantly higher. Pharmaceutical simulation of complex molecular systems, cryptographically relevant integer factoring, optimization problems at scales that outperform classical computing — most estimates place the requirement at hundreds to thousands of logical qubits, depending on the algorithm and the error rate. QuEra Computing's Libra system, targeting availability on Amazon Web Services in 2028, is designed for 256-plus error-corrected qubits — and that is already described as an early step. The DOE's Quantum Genesis program, which targets a fault-tolerant quantum computer by 2028, defines the early FTQC era as "low hundreds of logical qubits" — still dozens of times beyond LUMI-IQ's 9.

What 9 logical qubits enables is something different but genuinely valuable: the ability to run real fault-tolerant quantum circuits, test error correction under production conditions, develop the hybrid quantum-classical software stack that future larger systems will require, and generate foundational experimental data on how logical qubit systems behave at the intersection of AI and HPC workloads. LUMI-IQ Phase 3 is not a tool for solving drug discovery problems. It is the research platform on which Europe builds the expertise and infrastructure to eventually do so.

"Logical qubits make quantum calculations more reliable, while AI and HPC provide the scale, data processing and orchestration needed around them," said Kimmo Koski, Managing Director of CSC – IT Center for Science. "The LUMI AI Factory allows European researchers to develop innovative hybrid approaches in a world-class computing environment."

The 2029 completion of Halocene H5 aligns with IQM's broader roadmap target of fault-tolerant quantum computing by 2030. The LUMI-IQ platform is explicitly designed to scale from its 2029 foundation — the upgradeable architecture means the hardware does not need to be replaced to take advantage of future IQM QEC developments, including potentially the directional tile codes now under development.

Inside the LUMI Ecosystem

LUMI-IQ does not stand alone. The quantum system will be installed in the same Kajaani facility as LUMI-AI, the next-generation AI supercomputer for which EuroHPC signed a €387.8 million contract (approximately $451 million USD) with French infrastructure firm Bull on August 31, 2026. LUMI-AI will be built on AMD Instinct MI430X GPUs, delivering 288 teraFLOPS of hardware-based FP64 computing, with deployment scheduled for the second half of 2027.

Together, LUMI-AI and LUMI-IQ will form what CSC and IQM describe as a hybrid HPC-AI-quantum supercomputing platform: classical supercomputing for large-scale simulation and data processing, AI for optimizing complex workflows and analyzing data, and quantum processing for the specific computational tasks — optimization subproblems, quantum chemistry calculations, error-correction experiments — where quantum methods offer genuine advantages over classical alternatives. This tight integration is the architectural ambition that distinguishes LUMI's approach from standalone quantum computers or cloud-accessible systems: the quantum layer sits inside the same environment, on the same network, accessible through the same job-scheduling infrastructure, as one of Europe's most powerful AI systems.

"Owning an upgradeable quantum computer lets CSC and the LUMI AI Factory integrate the system into their own AI and HPC infrastructure," said Jan Goetz, CEO and Co-founder of IQM. "That's what lets a real ecosystem grow: researchers building on infrastructure they control, inside one of the most capable computing environments anywhere."

The site itself underscores the long-term infrastructure commitment. CSC's new data center in the Renforsin Ranta business park — on the site of a former UPM paper mill — has been under construction since January 2026, with facility completion targeted for spring 2027. It will house both LUMI-AI and LUMI-IQ when it opens, carrying on Kajaani's established tradition of repurposing industrial heritage for computational infrastructure. Like the original LUMI supercomputer, which is built on the same principle, waste heat from the new data center is designed to flow into Kajaani's district heating network.

The original LUMI, a project with a total budget of over €202 million (approximately $235 million USD), remains operational and continues to serve the AI Factory's current users while the new facility is completed.

IQM's Position as Europe's Quantum Hardware Company

The LUMI-IQ contract arrives at an inflection point for IQM. On July 2, 2026, the Espoo-headquartered company completed its business combination with Real Asset Acquisition Corp. and listed on Nasdaq under IQMX, becoming the first European quantum computing company listed on a major U.S. exchange. The listing closed with a pro forma cash position of €337 million (approximately $392 million USD), providing the runway to execute the Halocene roadmap through its 2029 completion target.

The LUMI-IQ selection marks IQM's second major superconducting deployment in a world-class supercomputing hub, following the Pathfinder installation at Oak Ridge National Laboratory — the home of Frontier, the world's most powerful open-science supercomputer — in June 2026. The two deployments together span the Atlantic and bracket IQM's commercial ambition: the Oak Ridge system demonstrated that IQM's on-premises model can operate inside the most demanding classical HPC environments on Earth; the LUMI-IQ contract commits that model to the fault-tolerant quantum era, on European soil, with national ownership.

How Does Distance-3 Error Correction Actually Work?

A distance-3 surface code encodes one logical qubit across a 3×3 grid of 9 data qubits, surrounded by 8 ancilla qubits used to measure "stabilizers" — products of Pauli operators acting on groups of neighboring data qubits. Every syndrome measurement cycle reads out these stabilizers; an error on any single data or ancilla qubit changes exactly which stabilizers show anomalies, telling the decoder where the error occurred. The decoder then determines the most likely correction without ever directly measuring the encoded quantum information. Google's 2023 Nature surface code study demonstrated this principle on superconducting hardware, showing logical error rates decreasing as code distance increased.

At distance-3, LUMI-IQ can protect against any single physical qubit error per syndrome cycle. At the architecturally supported distance-11, that protection extends to any combination of 5 simultaneous errors — dramatically more resilient, at the cost of 241 physical qubits per logical qubit instead of 17. The Halocene H5's 150 physical qubits can therefore support either approximately 8-9 logical qubits at distance-3, or a much smaller number at higher distances for experimental purposes.

Color codes, the second QEC approach LUMI-IQ will support, use a different lattice geometry (a 3-colorable tiling rather than a square grid) and have the useful property of supporting transversal implementation of the full Clifford group — the set of gates including Hadamard, CNOT, and phase gates — without magic state distillation. This makes them complementary to surface codes: researchers can choose the code whose gate properties best suit a given algorithm, rather than being locked into a single architecture.

Taken together, the combination of distance-3/distance-11 surface codes, color codes, lattice surgery, and T-gate distillation makes LUMI-IQ the most complete fault-tolerant quantum computing research platform in Europe — not because it can run the largest algorithms (9 logical qubits is far from that threshold), but because it implements the full toolbox that fault-tolerant quantum computing requires. Researchers who develop and validate algorithms on LUMI-IQ will be building on techniques that scale directly to the hundreds-of-logical-qubit systems the next generation of hardware will deliver.

"Already the first system delivered in 2027 will come with a state-of-the-art quantum processing unit with 150 qubits. This is only the beginning, however," said Mikael Johansson, Manager of Quantum Technologies at CSC. "Over the next years, LUMI-IQ will evolve to a fault-tolerant quantum computer through a series of upgrades that increase both qubit count and performance, making the LUMI AI Factory a world-leading European hybrid platform combining AI and quantum computing."

Currency conversions are approximate, based on exchange rates at time of publication.

Frequently Asked Questions

What is a logical qubit, and why does LUMI-IQ using them matter?

A logical qubit encodes one reliable unit of quantum information across many physical qubits, using continuous error detection to protect against the noise that destroys unprotected physical qubits within microseconds. The transition from physical to logical qubits is the central unsolved challenge of practical quantum computing — it is the difference between a quantum device that can run short demonstrations and one that can run algorithms long enough to be useful. LUMI-IQ will be the first superconducting quantum computer in Europe designed from the outset to sustain and operate on logical qubits, using distance-3 surface and color codes with lattice surgery and T-gate state distillation for universal fault-tolerant computation. For a deeper technical background on quantum error correction and fault tolerance, see TechTimes' earlier coverage of the field.

How many logical qubits does LUMI-IQ actually support, and is that enough for practical quantum computing?

LUMI-IQ Phase 3 (Halocene H5, 2029) will support up to 9 fault-tolerant logical qubits at distance-3. The architecture also includes headroom to encode logical qubits up to distance-11, enabling single-qubit high-fidelity experiments beyond the 9-qubit operational configuration. Nine logical qubits is not sufficient for the most commercially compelling quantum applications — drug discovery, materials simulation at scale, and optimization problems at useful size typically require hundreds to thousands of logical qubits. QuEra's Libra system, targeting availability on Amazon Web Services in 2028, is designed for more than 256 error-corrected logical qubits, and the DOE's Quantum Genesis program defines the early FTQC era as "low hundreds of logical qubits." LUMI-IQ's value lies in research: developing the algorithms, software stack, and operational expertise that future larger-scale fault-tolerant systems will require. It is a milestone on the path to quantum utility, not the destination.

What is the LUMI AI Factory, and how does LUMI-IQ fit into it?

The LUMI AI Factory is one of more than a dozen AI Factories established by the EuroHPC Joint Undertaking across Europe — dedicated compute ecosystems combining supercomputing capacity, curated data resources, and AI expertise for researchers, startups, and industrial users. LUMI-IQ is the quantum component of the LUMI AI Factory: it will be co-located with LUMI-AI, the next-generation AI supercomputer built on AMD Instinct MI430X GPUs, at CSC's new Kajaani data center in Finland. The combination is designed to enable hybrid workloads where quantum processors handle specific subroutines — optimization, error-correction experiments, quantum chemistry calculations — that classical GPUs handle inefficiently, while classical AI hardware manages the large-scale simulation and machine learning workloads quantum systems cannot yet run. LUMI-AI's AMD hardware selection and the €387.8 million contract with Bull are covered in TechTimes' prior reporting.

What makes IQM's approach to quantum error correction credible for the Halocene roadmap?

In June 2026, IQM researchers — working with Freie Universität Berlin, the University of Edinburgh, and Johannes Gutenberg-Universität Mainz — published research introducing directional tile codes: a new family of quantum low-density parity-check (QLDPC) error-correcting codes that demonstrated a 1,000-fold reduction in logical error rates compared to standard surface codes, at roughly 30 physical qubits per logical qubit, using only the nearest-neighbor gates that IQM's existing Crystal hardware already implements. The full announcement from IQM explains that Halocene H4 and H5 will deploy surface and color codes — not directional tile codes, which are still maturing — but the underlying research confirms that IQM's hardware is already architected for next-generation error correction. The upgrade path is grounded in demonstrated physics on production hardware, not solely in theoretical projections.