SUSE and Fujitsu announce a quantum management stack without naming a processor

A late 2026 product wraps NVIDIA's CUDA-Q software in the tools data centre teams already use. The release names no quantum chip, customer or price.

SUSE and Fujitsu have announced a quantum computing product that names no quantum processor. The 21 October release covers software that manages quantum jobs, due in late 2026.

The product is a management layer, not a quantum computer. SUSE packages, patches and schedules NVIDIA's quantum software on Fujitsu and Supermicro servers. A quantum job then runs like any other workload in the data centre.

The sales pitch centres on data sovereignty. The release says the stack keeps proprietary IP, algorithms and enterprise datasets strictly on premises or within private infrastructure. For a bank, insurer or drugmaker in Europe, data residency matters when buying the system, whatever the quantum hardware does.

A single support contract is the second argument. The release adds reference architectures validated in advance and one support contract covering the whole hardware and software stack. That fits SUSE's chosen position as a sovereign open infrastructure company. Data ownership and the single contract are also two of the release's three stated benefits.

The SUSE press release came from Berlin at 07:35 Eastern time. It describes an "AI-driven hybrid quantum platform". The platform combines SUSE AI Factory with NVIDIA and the NVIDIA CUDA-Q platform on Fujitsu servers. The release calls the partner "Fujitsu (FSAS)."

CUDA-Q is, in NVIDIA's words, an open source quantum development platform. It lets one program use GPU, CPU and quantum processor (QPU) resources. A QPU is the quantum chip itself.

What the software manages

SUSE AI Factory is the base platform, which SUSE announced in April at its SUSECON conference in Prague. The new part is the quantum extension on top. SUSE lists three features.

The first feature packages, deploys and manages CUDA-Q. SUSE says this is "eliminating manual patching and version-matching complexities."

The second is an "Open Source AI or Open Sovereign AI (OSAI) Building-Block Kit". It covers retrieval augmented generation, agentic coding and language model serving on Fujitsu and Supermicro hardware.

The third, a "Unified Hybrid Quantum Runtime," plugs CUDA-Q "directly into the compute broker" for "automated circuit submission and execution." A circuit is the program a quantum machine runs. The release does not define the compute broker.

All three features serve one idea. The release says FSAS treats "quantum circuits as workload extensions of SUSE AI Factory with NVIDIA." A circuit becomes a job that IT staff submit, patch and monitor with the tools they already use. The stated goal is "making AI-driven quantum computing manageable for standard IT teams."

The release names its users: system administrators, cloud architects and platform engineering teams. None is a quantum researcher. The product targets people who run servers, not people who write quantum algorithms.

SUSE's pitch is that a firm already running its tools can add CUDA-Q without a second management system. SUSE describes itself as "the only platform that manages any Linux and Kubernetes distribution." That is the company's own claim, taken from its boilerplate. The saving is in upkeep, not in quantum results.

All six items in the release are software operations. The three benefits are standardised IT operations, data ownership and a single support contract. One benefit promises to eliminate "the need for specialized quantum physics expertise." That is a claim about staffing, and the release gives no qubit count, performance figure, customer or price.

The vendors claim a first, at two different strengths. The subtitle promises "the first unified, production-ready hybrid quantum and AI stack on sovereign infrastructure." The body softens this to "one of the first times." Both describe the packaging of two software products, and neither claims a quantum result.

Simulation and the unanswered hardware questions

The one hint about the machine concerns simulation. The Fujitsu quote speaks of connecting "quantum simulation directly to production applications." Simulation means running quantum programs on ordinary computers instead of a quantum processor. CUDA-Q spans GPUs, CPUs and QPUs, so the same stack can run a simulator on ordinary GPU servers.

For a bank's quantum team, that simulator route is the practical product. Developers write one CUDA-Q program and test it on GPU servers the bank already owns. The Fujitsu quote supports this reading, saying customers can "supercharge their existing hardware investments."

A platform team weighing a 2027 quantum plan could use the product to rehearse CUDA-Q jobs on existing GPU servers under existing change control. Before signing, it should ask three questions in writing. Which processor or simulator runs the circuits? What does the support contract cover if a third party supplies the processor? What does the single contract cost? The release answers none of them.

Fujitsu's quote comes from Udo Wuertz, CTO Fsas Technologies EMEA. He mentions "our new quantum computer and Digital Annealer." The release does not say the new stack connects to either machine. It gives no qubit count for either.

The operations case still has merit. Thomas Di Giacomo, SUSE's chief technology and product officer, says teams can manage hybrid quantum workloads using "the enterprise tools, GitOps workflows, and security policies they already know." A team that avoids matching software versions by hand saves real effort. The release does not say what "AI-driven" means in practice, so the method behind that label is unstated.

The launch date and next conference

The release gives one date, "late 2026," and a list of software operations. That is enough to plan a software evaluation, but not a budget or a trial.

The release says the joint solution will be available in "late 2026." The SUSECON 2027 page puts the next conference in Barcelona from 15 to 17 February 2027. The release gives a date, and SUSE's next conference falls soon after it.

The Barcelona conference falls after the product is due, so it is the first SUSE conference after the promised launch. If SUSE's demonstrations there still run on GPU simulation, the product is a workload manager for existing servers. A named processor on stage would turn the same sovereignty pitch into a quantum offer.