Hewlett Packard Enterprise (HPE) launched the Quantum Scaling Alliance, a consortium focused on building a practical quantum supercomputer by tightly coupling quantum hardware with classical high-performance computing (HPC) and advanced networking. The initiative brings together organizations spanning semiconductor engineering, quantum control, error correction, and full-stack system integration.

Masoud Mohseni, who leads the quantum team at HPE Labs and serves as quantum system architect for the effort, said the alliance was created in response to a growing realization that much of today’s quantum ecosystem does not scale.

“Many of the components and elements have matured enough to be integrated at this moment, but they will not scale well in the vertical approaches people have been pursuing,” Mohseni said. “You need to bring the best in class of every part of the stack.”

The group brings together HPE, Applied Materials, Qolab, Quantum Machines, Riverlane, Synopsys, 1QBit, and researchers at the University of Wisconsin. It combines semiconductor fabrication capabilities, quantum control systems, error-correction expertise, EDA and simulation tooling, and academic benchmarking into a coordinated effort to build a hybrid architecture across the classical and quantum stacks.

Mohseni told SDxCentral that horizontal integration allows teams with deep domain expertise to build each layer of the system.

“A modular structure with very fast HPC interconnects lets you do real-time controller error correction and workload management across many GPUs, which is not possible today,” Mohseni said. “New players might come in, and you can absorb them, or you can let go of technologies that did not work out. In a vertical integration, it is harder. People double down even if it is not landing well.”

A major focus is real-time error correction, which Mohseni called a defining challenge for large-scale systems.

“The biggest problem in building a quantum computer is noise, which is known as decoherence,” Mohseni said. “You have to correct errors really fast, so it has to happen in real time.”

That requires rapid data movement between quantum processors and classical accelerators and constant recalibration as systems drift.

“You might have a system that works for a while but then gets noisy in a way that your control mechanisms get outdated, so you have to keep updating your quantum control,” Mohseni added.

The alliance aims to demonstrate full error-corrected pipelines operating across hundreds of logical qubits, which Mohseni described as a key inflection point.

“When you have interactions among hundreds of logical qubits and a full error-correction pipeline, that is when we are leaving the academic era and getting into industry-level runtime,” Mohseni said.

Application priorities include quantum simulation, quantum chemistry, and quantum-generated datasets for AI pipelines. Mohseni said the team is focused on inherently quantum problems.

“You need a super-polynomial speed-up for the advantage to land in real value,” Mohseni said, adding that recent algorithmic progress has brought some workloads “under 10 million qubits,” increasing pressure to solve system-integration challenges.

The alliance is co-led by Nobel laureate John Martinis, now cofounder and CTO at Qolab. Mohseni, who worked with him for more than a decade, credited Martinis with helping pioneer early scalable superconducting designs.

Mohseni added that the timing of the alliance launch reflects progress across hardware and algorithms.

“Qubits are much better, algorithms have matured, and semiconductor manufacturing allows you to build much larger integration,” he said. “When you put it all together, the possibility is now real.”