IQM puts uptime, scheduling at the heart of quantum-HPC integration

IQM has sold 30 quantum computers and delivered roughly 20, including 10 integrated with high-performance computing systems, co-founder Juha Vartiainen said at the Quantum + AI Tech Show in London on November 4. Vartiainen said an IQM system at Germany's Leibniz Supercomputing Center (LRZ) ran more than 200 days with high uptime and stable performance, and an IQM/LRZ case study of a 20-qubit superconducting machine found scheduler-controlled automatic recalibration is needed to keep the system available, with quick recalibration taking about 40 minutes and a fuller procedure about 100 minutes. The deployment account notes a warm-up event can force a two-to-five-day cooldown plus recalibration, making uptime and workload scheduling central to quantum-HPC integration.

IQM has integrated quantum computers into 10 data centers, and its experience at Germany’s Leibniz Supercomputing Center LRZ underscores the need for reliable performance and shared workload management, according to company co-founder Juha Vartiainen. Speaking at the Quantum + AI Tech Show in London, on November 4, Vartiainen said IQM had sold 30 quantum computers and delivered approximately 20, including 10 integrated with high-performance computing HPC systems. According to Vartiainen, an LRZ customer system had been operating for more than 200 days with high uptime. “The performance has stayed quite constant over that time,” he added. His comments place operational measures alongside processor specifications as quantum computers become resources that computing centers must maintain, schedule, and make available to users. Calibration becomes a scheduling task A published case study by IQM and LRZ researchers https://arxiv.org/abs/2509.12949 documents integrating a 20-qubit superconducting quantum computer into LRZ’s HPC infrastructure. The authors describe a co-located, loosely integrated model, providing practical experience for future systems with tighter quantum-classical connections. Unlike conventional HPC nodes, quantum processors have characteristics that change during operation, requiring regular recalibration. The study identifies automatic calibration, controlled by the HPC scheduler, as necessary to maintain availability while accommodating users’ workloads. An Open Compute Project account of the deployment https://www.opencompute.org/news/toward-a-blueprint-for-quantum-supercomputer-co-deployments-iqms-20-qubit-superconducting-quantum-computer-integration-into-the-hpc-infrastructure-at-lrz/ reports that a quick recalibration takes approximately 40 minutes, while a fuller procedure takes around 100 minutes. Scheduling those operations makes calibration part of resource management rather than a separate maintenance activity. The LRZ deployment used standard 1 Gb/s Ethernet, with latency proving a more important consideration than bandwidth, according to the case study. Keeping the system available The facility itself needs to support consistent operation. The deployment work included assessing magnetic interference, vibration, temperature, and humidity before selecting the installation site. Redundant power and cooling are particularly important. A disruption that allows the quantum processor to warm up can require a cooldown lasting two to five days, followed by recalibration, according to the deployment account. Those requirements extend the integration task beyond connecting a processor to a classical computer. Operators must coordinate facilities, monitoring, maintenance, and software access to provide a usable computing resource. IQM’s deployments also include Pathfinder at Oak Ridge National Laboratory https://www.businesswire.com/news/home/20260616373175/en/IQM-Deploys-Its-First-U.S.-Quantum-Computer-at-Oak-Ridge-National-Laboratory , a 20-qubit system connected to the laboratory’s HPC testbed. Oak Ridge owns and operates the machine on its campus. Shared access introduces another operational requirement. In a separate panel at the London event, BT’s head of optical research and quantum communications, Andrew Lord, discussed customers accessing quantum computing for algorithms that require only a fraction of a machine’s available time. Rather than each customer owning a computer, that model would require providers to allocate access across multiple users. Lord identified scheduling and orchestration as necessary to ensure customers receive the computing resources they need.