HPE combines Cray supercomputers with quantum processors: a new stage in hybrid computing
Hewlett Packard Enterprise (HPE) is making a decisive move toward hybrid computing, expanding its partner ecosystem to integrate its Cray supercomputers with quantum technologies. The alliance includes key industry players: Intel, IQM, Qblox, Quantinuum, QuEra Computing, Quantum Machines, Rigetti, and Riverlane. This is not just a list of big names — it is a strategic assembly of all necessary components to build viable hybrid systems.
Architecture of the Future: Supercomputers + Quantum Processors
The essence of the initiative lies in directly connecting the high-performance HPE Cray platform with quantum processing units (QPUs), qubit control systems, and error correction modules. As part of the project, partners will deploy specialized testbeds. These testbeds will serve as a proving ground for debugging hybrid algorithms, verifying software stack compatibility, and objectively assessing performance in real-world scenarios.
Key point: quantum computing alone will not replace classical supercomputers. Their strength lies in solving a narrow class of problems where classical machines are powerless (e.g., molecular modeling or optimizing complex supply chains). HPE is betting on symbiosis: the classical Cray handles 99% of computations, while the quantum module is engaged to process critical subtasks.
A Practical Perspective on the Problem
For now, quantum processors remain "raw" — with high error rates and limited coherence times. This is why the involvement of Riverlane (error correction) and Quantum Machines (control systems) is critically important. Without them, a hybrid system would be nothing more than a laboratory experiment. HPE is essentially trying to create a "quantum accelerator" analogous to a GPU, but orders of magnitude more complex.
My assessment: this alliance is one of the most pragmatic approaches to quantum computing I have seen. Instead of waiting for a perfect quantum computer, HPE is building the infrastructure for its gradual integration. If the testbeds prove effective, we could see the first commercial hybrid solutions as early as 2026-2027. However, the main risk is scalability: so far, none of the partners have demonstrated stable operation with hundreds of logical qubits.