Doudna supercomputer from Dell and NVIDIA set to accelerate US scientific discovery and AI innovation

Dell and NVIDIA’s Doudna Supercomputer Set to Advance US Scientific Research

The US Department of Energy (DOE) is preparing to launch its newest high-performance computing system, the Doudna supercomputer, at the National Energy Research Scientific Computing Center (NERSC) in 2026. Named after Jennifer Doudna, the Nobel Prize-winning biochemist behind CRISPR gene-editing, this system represents a significant upgrade for scientific computing, integrating powerful AI capabilities with traditional simulation tasks.

Unmatched Performance for Complex Scientific Challenges

Doudna will deliver more than ten times the performance of its predecessor, Perlmutter. Built by Dell Technologies and powered by NVIDIA’s latest Vera Rubin platform, the system uses Dell’s liquid-cooled ORv3 servers to handle demanding workloads more efficiently. This leap in computing power will enable researchers to run complex simulations and data analyses much faster, accelerating progress across multiple scientific fields.

The supercomputer is optimized for tasks such as molecular dynamics, high-energy physics simulations, and large-scale AI training. By combining simulation, data analysis, and AI into a unified environment, Doudna facilitates multidisciplinary research workflows that require high efficiency and scalability.

Infrastructure Built for Versatility and Speed

Doudna’s architecture features Dell Integrated Rack Scalable Systems equipped with PowerEdge servers and NVIDIA accelerators. The NVIDIA Quantum-X800 InfiniBand network provides high-speed communication among thousands of compute nodes, essential for data-intensive operations.

Storage is handled through a parallel file system with smart data tiering, offering quick, real-time access to vast datasets. The system supports a heterogeneous workflow environment, allowing it to adapt to various research demands—from traditional simulations to AI-based inference and training.

Driving Advances in AI, Quantum, and Energy Research

A key strength of Doudna is its capacity to support AI-driven scientific work at scale. Researchers will be able to embed machine learning models directly into simulation pipelines, enhancing precision and reducing computation time in areas like materials science, climate modeling, and biomolecular design.

Additionally, Doudna will facilitate quantum computing research via NVIDIA’s CUDA-Q platform, enabling the simulation and development of quantum algorithms. This positions the system as a testing ground for hybrid quantum-HPC technologies.

Importantly, Doudna will contribute to fusion energy research by providing the computational power necessary to simulate plasma physics and reactor designs, potentially accelerating progress toward practical fusion power.

Enabling Nationwide Collaborative Research

Integrated with the Energy Sciences Network (ESnet), Doudna will allow researchers across the country to stream experimental and observational data directly into the system. This real-time data access improves responsiveness to new findings and supports interactive workflows.

With over 11,000 users depending on NERSC’s resources, preparations are in place to ensure smooth adoption of Doudna’s capabilities. The system will support AI-driven automation and interactive computing environments suited for data-intensive research.

Setting a New Standard for Scientific Computing

The arrival of the Doudna supercomputer marks an important step in advancing US scientific infrastructure. By combining resources from Dell Technologies, NVIDIA, and the DOE, the system is positioned to accelerate discovery across a wide range of disciplines, from fundamental physics to next-generation energy solutions.

Researchers seeking to deepen their expertise in AI and computational science may find valuable resources at Complete AI Training, which offers courses designed for professionals in science and research fields.