Breaking Update: Here’s a clear explanation of the latest developments related to Breaking News:NASA’s Athena supercomputer comes online– What Just Happened and why it matters right now.
NASA’s Athena supercomputer has come online at the agency’s Modular Supercomputing Facility at the Ames Research Center in California.
Delivering 20.13 petaflops of compute power, Athena forms part of the space agency’s High-End Computing Capability project and replaces the 7.09 petaflops Pleiades, which was decommissioned earlier this month.
According to NASA, Athena is comprised of 1,024 AMD Epyc Turin nodes and 786TB of memory. The system will be made available to NASA researchers and external scientists and researchers supporting NASA programs who can apply for time to use the system.
“Exploration has always driven NASA to the edge of what’s computationally possible,” said Kevin Murphy, NASA’s chief science data officer. “Now with Athena, NASA will expand its efforts to provide tailored computing resources that meet the evolving needs of its missions.”
The CPU-based Athena is one of a number of supercomputers owned by NASA, with the rest of its fleet comprising the 20.67 petaflops Cabeus; the 13.12 petaflops Aitken; the 8.32 petaflops Electra; the 8.1 petaflops Discover; and the 154.8 teraflops Endeavour.
However, in 2024, a special report conducted by the NASA Office of Inspector General (OIG) has found that the space agency’s missions are being hampered by inadequate high-end computing (HEC) resources, with its systems largely relying on CPUs, thus making them insufficient for the agency’s tasks.
The report also said that NASA’s HEC resources are “oversubscribed and overburdened,” meaning that Mission Directorates are requesting more computing time than existing capacity can provide. Consequently, the OIG proposed several recommendations to the space agency, which included establishing an executive leadership team to help reposition NASA’s HEC resources so they meet the agency’s “specialized needs.”
In March 2025, NASA added 350 Nvidia GH200 nodes to Cabeus, increasing its compute power by more than 13 petaflops and doubling its total peak capability.