A proposed University of Michigan computing center will support research related to nuclear weapons and plutonium pit development for Los Alamos National Laboratory, a lab spokesperson told the Exchange Monitor in an email.
“Los Alamos is partnering with the University of Michigan on a high-performance computing center,” the spokesperson said, adding it would not be a “commercial data center” but instead a computing facility for researchers from the lab and University of Michigan. “Los Alamos National Laboratory uses high-performance computing to run advanced simulations across a wide range of scientific fields, including materials science, biological systems, astrophysics, energy research and national security.”
While the facility will use physics-based simulations to “study weapon performance, aging, and lifecycle processes,” plutonium pits will not be developed at the center, the spokesperson emphasized.
The spokesperson added that “a core part of the laboratory’s mission is stockpile readiness science, which ensures the safety, security and reliability of the U.S. nuclear stockpile. The computing facility at the University of Michigan will strengthen the laboratory’s ability to advance and sustain its national security missions.”
Classified and unclassified research will take place at the facility, meaning University of Michigan students and faculty can partake in the unclassified research and partner with Los Alamos, the spokesperson said.
The facility would be in Ypsilanti Township, an apparent point of contention among the Township’s residents. The Michigan Daily said Ypsilanti residents and University members worry about the effects to the environment and electrical grid, and have also expressed concerns over the potential that the center could be used to develop nuclear weapons, which the University has denied.
DOE has selected four nuclear complex sites to be used for new artificial intelligence (AI) data centers, and has been looking to invite private sector partners to work to build AI data centers and energy generation projects. DOE has said in the past the push for AI centers on federal lands took after the executive orders, Accelerating Federal Permitting of Data Center Infrastructure, Deploying Advanced Nuclear Reactor Technologies for National Security and Unleashing American Energy.
In an interview on the sidelines of Exchange Monitor’s Nuclear Deterrence Summit in late January, Thom Mason, Los Alamos’s lab director, told the Monitor that AI large language models, which require a lot of power, are helpful for the lab to use in a classified setting. Historically, data on the nation’s nuclear weapons tests is “voluminous and hard for any single person to kind of absorb,” Mason said. He added that “using researchers and private sector resources “around the world” has been helpful to “fill the gaps in the training data” on open source AI tools “that exist because we don’t share classified information.”