Classical computing leaders move into quantum: IQM and AMD run a hybrid chemistry workflow

The companies that lead classical computing are moving into quantum. In the past few weeks AMD has shown hybrid quantum work with our portfolio company IQM and with IBM, and Intel has agreed to buy quantum software company Classiq. NVIDIA has spent the past year building its own platform for linking quantum processors to GPUs. The IQM and AMD demonstration is a small, carefully scoped example of what this engagement looks like in practice.

What IQM and AMD did

The demonstration uses Quantum-Selected Configuration Interaction (QSCI) to model the hydrogen molecule, H₂, in the STO-3G basis. The IQM Garnet quantum computer, accessed through IQM’s cloud platform IQM Resonance, samples electron configurations. An AMD GPU running the ROCm software stack then filters out unphysical samples, builds the Hamiltonian matrix over the remaining configurations and diagonalises it using JAX.

The teams ran the workflow first on a simulator and then on real hardware. All 17 bond lengths on the H₂ dissociation curve were measured on the IQM quantum computer, and the results were checked against an exact classical solution.

What it does not show

IQM is explicit that this demonstration does not show quantum advantage and was not designed to. H₂ is small enough to solve exactly on a classical computer. That is what makes it a useful benchmark: every step of the pipeline can be checked against a known answer.

Classical computing leaders step up

For AMD, the IQM work is one of two recent quantum moves. On 5 November, IBM, AMD and the Quantum Science Center at Oak Ridge National Laboratory launched a co-design platform for quantum error correction that places IBM’s 120-qubit Nighthawk processor alongside Umbra, a system built on AMD Instinct GPUs and EPYC CPUs.

NVIDIA has been building in the same direction for longer. Its NVQLink architecture, announced in October 2025, connects quantum processors to GPU supercomputers through the CUDA-Q software platform, and launched with 17 quantum hardware builders, IQM among them.

Intel has taken the most direct step. On 26 October it agreed to acquire Classiq, whose platform automates the design of quantum algorithms across many types of hardware, and plans to combine it with its own silicon spin-qubit and manufacturing work to become a full-stack quantum provider. Intel’s quarterly filing puts the price at about $1.9 billion, plus up to $350 million tied to milestones. The deal is expected to close in the first half of 2027, subject to regulatory approval.

As Owen Lozman told EE Times, Intel’s manufacturing and packaging are real strengths, but scaling spin qubits still involves many engineering challenges, even if the physics is right. Buying a software layer it did not previously own lets Intel move early while that hardware matures. The harder part will be keeping the platform open to other companies’ hardware, which, in Owen’s words, will take a lot of discipline.

Why it matters

Taken together, these moves point the same way. Classical computing leaders increasingly treat the quantum processor as another accelerator alongside CPUs and GPUs, and they are investing in the software, interconnects and partnerships needed to use one. Useful quantum computing will come from systems in which quantum hardware and high-performance computing each do the part they are best at. Making that division of labour work reliably, end to end, is engineering that has to be done before any claim of advantage can be tested properly.

For quantum companies, this brings more capable partners and more demanding ones. A small, well-defined problem, a checked answer and a clear statement of scope, as in the IQM and AMD work, give researchers, customers and partners something they can evaluate and build on. That is the kind of progress that compounds, and it reflects our ambition to help build globally leading quantum companies.

What comes next

IQM and AMD say the next steps are larger molecules, better circuits, more shots and measuring how the GPU contribution changes with problem size. They expect the method to start paying off at around eight qubits and above; that remains to be demonstrated. Today the two systems connect through the cloud, and the teams are working toward tighter integration between GPU nodes and quantum processors.

Congratulations to the IQM and AMD teams. Read IQM’s full account of the demonstration.