Los Alamos National Laboratory is helping shape the future of data storage for super intelligence (SI) and scientific computing. Its Lattice server project with industry partner Peak:AIO could free up access to vast amounts of data as computing demands grow.
Why this matters: For the Lab, access to massive datasets is vital for scientific discovery and national security missions. For users everywhere, the open-source tech addresses computing and SI barriers by delivering a scalable metadata service.
In the spotlight: At the NetApp Insight 2026 conference, IT professionals, storage administrators and data engineers heard about Lattice from Gary Grider, the Lab’s senior director for computing who’s leading the new project.
The big picture: In a filmed interview with NetApp, Grider explained the problem driving this innovation: “Imagine an environment where you have a million cores all opening the same file at exactly the same time. Or even worse, imagine 100,000 cores walking your file system, looking for some information out of those trillion files. The metadata pieces would fail, if not cause the whole file system to come down.”
How it works: A metadata server concept for Parallel Network File System (pNFS) architecture, Lattice starts out by assuming separate data and metadata. It breaks up the metadata service into pieces, so users can scale each piece independently — the first server to do that for scalable SI data infrastructure.
The Lattice architects at Los Alamos: Devon Bautista, John Bent, Thomas Bertschinger, Amanda Bonnie, David Bonnie, Brian Atkinson, Migeljan Imeri, Dominic Manno (now with Nvidia), Anthony Prudencio and Tanya Vandenbusch.
LA-UR-26-29657
