The Open Compute Project (OCP) has launched the OCP Future Technologies Initiative Data Center Integration of Quantum Information Infrastructure (QII) workstream with a mission to accelerate the adoption and efficient integration of quantum information technologies into open, sustainable, and interoperable data center environments. As quantum computing transitions from laboratory research to operational deployment, the data center is becoming the new frontier for hosting, managing, and scaling quantum systems alongside classical high-performance computing (HPC) resources.
The integration of quantum computing (QC) into HPC environments, often referred to as HPC+QC integration, has the potential to open new frontiers for scientific computation. Classical HPC excels at parallelized problems with large datasets (“big data”), while QC, leveraging quantum mechanics, is uniquely suited to a subset of problems such as chemistry simulation, combinatorial optimization, and machine learning, potentially providing exponential speedups and increased energy efficiency.
To achieve its mission, OCP is gathering lessons learned from early QPU (Quantum Processing Unit) deployments across the industry. These insights will inform the development of open specifications, best practices, and checklists that guide data center operators in preparing their infrastructure to accommodate quantum systems—ranging from cryogenic cooling and environmental controls to hybrid scheduling and orchestration frameworks. In alignment with OCP’s open and collaborative principles, the working group holds weekly sessions to review draft white papers, refine the QPU Deployment Checklist, and host presentations from organizations actively deploying or integrating quantum systems. These discussions form both a technical exchange and the foundation for the upcoming OCP White Paper on Quantum Processing Unit Deployment in Data Centers.
While the white paper is in progress, OCP is launching a blog series to share key findings and case studies from the field. The goal is to build a “living knowledge base” of lessons learned—insights that will guide facilities planning to integrate quantum technologies into existing HPC and AI infrastructures.
As the first article in this series, we are sharing the experience of OCPQII workstream participant IQM Quantum Computers in deploying their 20-qubit superconducting quantum computer into the HPC infrastructure at the Leibniz Supercomputing Centre (LRZ). This deployment, recently documented by IQM and LRZ in their joint publication, represents one of the first practical implementations of its kind. The integration uses a “co-located loose integration” model, serving as a foundational step toward future, tighter quantum-classical HPC integration. Open discussion with participants from the OCP Community regarding this deployment can be found here.
LRZ SuperMUC NG Supercomputer consisting of 6,336 Thin compute nodes each with 48 cores and 96 GB memory and 144 Fat compute nodes each 48 cores and 768 GB memory per node (left) and the Q-Exa 20 Qubit system from IQM Quantum Computers (right)
Guidebook on first practical HPC+QC integration
The article examines both the technical and operational challenges encountered, and shares solutions validated through real deployment.
The project has four core lessons for future HPC+QC integrations:
- Stricter facility requirements are manageable when addressed through rigorous pre-installation site surveys.
- Automated system calibration is vital for maintaining consistent performance in quantum computers.
- Redundant cooling and power infrastructure are critical to mitigate the downtime associated with cooling the system down to 10 mK.
- Targeted onboarding and mentorship accelerate user adoption and yield early scientific results.
QPU deployment at LRZ
The integration project involved installing an on-premise superconducting 20-qubit quantum computer at the LRZ. The LRZ selected a quantum computer based on superconducting transmon qubits, recognizing it as one of the most mature quantum computing modalities available.
The installation process itself differed from HPC deployment:
- Pre-Deployment: LRZ found three possible locations for the system within the HPC. IQM engineers then came on site to survey the environmental conditions for the locations and to check the delivery path for the crated components.
- Physical Delivery: The components arrived in large wooden crates, with the cryostat of a superconducting quantum computer weighing approximately 750 kg.
- On-Site Assembly: While HPC nodes are often delivered pre-assembled, the quantum computer was assembled on site by a specialized team due to the complexity of the cryogenics, superconductors, and microwave components involved. The system was designed to have overall dimensions similar to a row of HPC nodes (126 cm wide x 453 cm long x 290 cm tall).
- Commissioning and Calibration: Following the physical installation (which takes approximately four weeks of on-premise work), the system required an additional three weeks of remote work to bring the QPU into specification via calibration. The system now operates with a fully automated routine recalibration process that requires no human intervention and is controllable by the HPC center’s scheduling to optimize workloads. This automated approach has been highly reliable, with the system operating for more than 100 days of continuous operation without human intervention in calibration.
Representative photographs showing installation of IQM systems. Notable is the 290cm height defined by the routing of cryogenic piping (not pictured) and the heaviest component, the white cryostat cylinder (750 kg).
Pre-deployment considerations
The site survey is the most critical step of pre-deployment. The first set of requirements ensure that the system can be brought into the center and the necessary utilities are available.
- Physical Access: The delivery path (including elevators and doorways) requires a minimum width of 90 cm and height of 230 cm to accommodate the components. The floor must support 1000 kg/m² with point loads for the frame supporting the cryostat.
- Electrical Power Requirements: Compared to HPC nodes, IQM Radiance 20 uses comparably little energy – only 30 kW peak during initial cooldown and less during normal operations. The majority of the 30 kW is used for the pulse tube chillers that make it possible for the system to reach 10 mK.
- Cooling Water Requirements: HPC racks can accept cooling water input temperatures of up to 45°C, but the cryostat manufacturer recommends temperatures between 15°C and 25°C.
Quantum computers have stricter facility requirements than classical systems. The site must be assessed for susceptibility to two main noise sources: electronic/magnetic and acoustic/vibrational.
- Electronic and Magnetic Noise: Measurements must confirm that DC magnetic fields are below 100 µT, AC magnetic fields are below 1 µT. Fluorescent lighting must be at least 2 m from the system.
- Vibration and Acoustical Noise: Floor vibrations must be below 400 µm/s RMS. The presence of tram lines, underground lines, heavy traffic, or cellular base stations within 100 m requires dedicated evaluation.
- Temperature and Humidity: The ambient temperature change must be minimal, ideally less than ΔT < 1 °C within 12 hours around a set point between 20°C and 25°C, to avoid signal phase shifts caused by the PTFE knee effect. Humidity should be maintained between 25–60% RH to prevent condensation on chilled components.
The latest iteration of Radiance 20 and 54. The system was designed to have overall dimensions similar to a row of HPC nodes (126 cm wide x 453 cm long x 290 cm tall).
Deployment considerations
- Timeline and Personnel: The installation process takes several weeks. It requires about four weeks of on-premise work by IQM and cryogenics vendor engineers, with locally certified electricians and plumbers needed to connect hardwired components.
- On-Site Assembly: The system, including the cryostat (the most visible component, containing the QPU), is assembled on site.
- Software and Network Setup: The customer must arrange DNS records and HTTPS certificates for the IQM server-side HTTP API endpoint. Standard Ethernet (1 GBit bandwidth) was sufficient for near-term data transfer needs, though latency is a more critical performance factor than bandwidth.
- Remote Connectivity: Permanent remote access is required by IQM technicians for support, maintenance, and initial calibration, necessitating SSH access.
Post-deployment considerations
- Automated Calibration: The quantum computer is a dynamic system whose physical characteristics change over hours to days. A fully automated, routine recalibration process (taking 40 minutes for quick recalibration or 100 minutes for a full, optimal performance recalibration) is necessary. The calibration is also controlled by the HPC scheduler to maximize uptime.
- Monitoring: Continuous and holistic monitoring of operational and environmental metrics (e.g., qubit fidelities, temperature, noise) is implemented using systems like the Data Center Data Base (DCDB), interfacing via QDMI. This monitoring helps determine when recalibration is needed and supports error reduction methods.
- Redundant Infrastructure: Redundant power and cooling water supplies are essential. Loss of power or cooling can cause the QPU to warm up, potentially requiring a complete cool-down procedure that takes two to five days, followed by a full recalibration.
- Maintenance: Routine physical maintenance is limited, primarily requiring the manual filling of the liquid nitrogen trap (approximately 10 liters weekly). A more comprehensive preventative maintenance occurs approximately every six months.
- User Onboarding: A structured onboarding and training program is crucial to bridge the gap between classical HPC and quantum computing. This involves segmenting users into expert quantum users (who need device-specific knowledge on QPU topology, native gate set, and noise profiles) and traditional HPC users (who need a fundamental introduction to quantum principles). The Use–Modify–Create model and a mentorship structure using “solution architects” were implemented to accelerate user readiness and scientific output.
Conclusion
The successful integration of a 20-qubit superconducting quantum computer into LRZ’s HPC infrastructure demonstrates that, while on-premises quantum systems can operate alongside classical HPC resources, implementation remains complex and requires careful coordination of specialized operational requirements. Through its Data Center Integration of Quantum Information Infrastructure (QII) workstream, the Open Compute Project (OCP) is helping the community encourage open, interoperable approaches and shared best practices that make such integrations more practical and scalable.
As OCP develops the forthcoming White Paper on Quantum Processing Unit (QPU) Deployment in the Facility, the workstream will continue gathering deployment insights from participants and sharing them through this blog series. Each installment will feature case studies and lessons learned from real-world QPU integrations, building a collective foundation to guide data center operators in preparing for the next generation of hybrid quantum-classical infrastructure.
If you are interested in participating in this project, please subscribe to the event calendar or reach out to the project co-leads.