The National Energy Research Scientific Computing Center (NERSC) at Lawrence Berkeley National Laboratory has taken a major step toward deploying its next-generation supercomputer by receiving a smaller pilot system called Cech in January 2026. This early-access system allows NERSC staff and Dell Technologies to test assembly, installation, and integration processes before the full-scale Doudna supercomputer arrives in late 2026, ensuring a smooth deployment for the facility's 11,000 users across national laboratories and universities. The naming of these systems reflects a lineage of scientific discovery. Doudna is named after Jennifer Doudna, the biochemist who won the Nobel Prize for developing CRISPR gene-editing technology. Cech honors Thomas Cech, a Nobel Prize-winning chemist whose groundbreaking work on RNA catalysis in 1982 laid the foundation for Doudna's later research. What Makes Cech Different From Previous Supercomputers? Cech represents a significant leap in how supercomputers balance raw computing power with energy efficiency. The system packs 72 NVIDIA Grace CPUs (central processing units) and 144 NVIDIA Blackwell GPUs (graphics processing units) into Dell's energy-efficient IR7000-series direct liquid-cooled racks. This combination delivers 5.76 petaflop/s of FP64 computing power for traditional simulation work and 1.44 exaflop/s of NVFP4 performance for AI-intensive tasks, making it a powerful system in its own right even before the full Doudna deployment. To put this in perspective, a single cabinet of NVIDIA Grace-Blackwell nodes has as much FP64 computing capability as twice the entire Edison system, which was NERSC's flagship supercomputer from 2014 through 2019 and filled an entire data-center floor. This dramatic improvement in density and efficiency represents how far computing technology has advanced in just over a decade. How Is Cech Solving the Data Center Cooling Challenge? One of the most innovative aspects of Cech is its approach to managing heat generated by powerful GPUs. The system features several cutting-edge cooling technologies designed to reduce operational costs and environmental impact. These include: - Low-latency networking: NVIDIA Quantum-X800 InfiniBand interconnects allow GPUs to communicate at extremely high speeds with minimal delay, reducing the time data spends in transit and generating less heat overall. - Advanced coolant distribution: Rittal's in-row V3.5 coolant distribution unit represents the first U.S. deployment of this technology, enabling more efficient heat removal directly from the source. - Rear-door heat exchange: Dell's PowerCool Enclosed Rear Door Heat Exchanger (eRDHx) is engineered to reduce cooling costs by 60% compared to traditional cooling methods, a substantial improvement for facilities running 24/7 operations. "NERSC has a long history of deploying first-of-a-kind systems while pushing the boundaries of energy-efficient computing, and Cech is the latest example of this," explained Brian Friesen, NERSC-10 Deputy Director for Systems. "Integrating these cutting-edge NVIDIA AI infrastructure with Dell's pioneering liquid-cooling technology gives NERSC staff an incredible sandbox to test this powerful combination, and ensure a seamless, sustainable deployment for Doudna". What Will Cech Be Used For During Its Testing Phase? Rather than immediately opening Cech to the broader scientific community, NERSC is using this early-access period for intensive preparation work. The facility is benchmarking the system's performance and testing new software infrastructure needed to manage Doudna when it arrives. This includes developing and refining a complex modular software stack based on Omnia and OpenCHAMI, which will orchestrate the supercomputer and manage its user software environment. NERSC and Dell are also collaborating on monitoring, telemetry, and alerting software infrastructure that will support both Doudna and future systems. Additionally, the team is working to make decades of data stored on NERSC's Community File System and HPSS archive accessible on Doudna, and to establish connections to external resources through ESnet, the Department of Energy's dedicated network for science. After this initial development period, Cech will support the Genesis Mission, the Department of Energy's artificial intelligence initiative, providing researchers with access to a powerful system for AI-intensive workloads. Why Does This Matter for the Future of Scientific Computing? The deployment of Cech represents a broader shift in how federal agencies approach high-performance computing and artificial intelligence infrastructure. By testing new technologies and processes at scale before full deployment, NERSC is reducing the risk of costly integration problems and ensuring that Doudna will be ready to support groundbreaking research from day one. "Getting Cech in the data center now is essential to ensure the successful deployment of Doudna coming up at the end of the year," said Rollin Thomas, NERSC-10 project director. The innovations being tested on Cech, particularly the combination of advanced cooling technology and GPU acceleration, are likely to influence how other federal agencies and private companies design their own data centers. As AI workloads continue to grow in complexity and scale, the ability to pack more computing power into smaller spaces while reducing energy consumption becomes increasingly important for both cost management and environmental sustainability. NERSC Director Sudip Dosanjh emphasized the significance of this moment: "We are excited to take delivery of the early access system. This is an important step in deploying Doudna for NERSC's 11,000-plus users. We plan to use the EAS to develop and test the software stack for Doudna in collaboration with Dell Technologies, NVIDIA, and our other partners. This opportunity to polish our processes helps ensure a great user experience when we deploy the full system". When Doudna arrives in late 2026, it will serve as a blueprint for how federal agencies scale high-performance computing and AI infrastructure securely, efficiently, and at mission speed, according to Paul Perez, senior fellow at Dell Technologies Federal. The lessons learned from Cech's deployment will help ensure that transition is as smooth as possible.
