NASA unveils 500× radiation-hardened AI chip
- NASA and Jet Propulsion Laboratory said on May 12, 2026, that their next-generation High Performance Spaceflight Computing processor had entered testing. (nasa.gov) - The key figure is “over 100 times” current space-processor computing capability, with Microchip Technology developing the chip under NASA’s High-Performance Spaceflight Computing program. (nasa.gov) - NASA says JPL is running radiation, thermal, shock and functional tests as the processor moves toward future lunar and planetary missions. (jpl.nasa.gov)
NASA’s Jet Propulsion Laboratory said on May 12 that its next-generation High Performance Spaceflight Computing processor is undergoing testing, marking NASA’s latest public update on a project aimed at giving spacecraft far more onboard computing power. The processor is designed to survive deep space conditions while handling artificial intelligence, autonomy and other demanding workloads that existing radiation-hardened space computers struggle to support. (nasa.gov) NASA and Microchip Technology are developing the processor through a public-private partnership first announced in 2022. (nasa.gov) NASA says the chip is intended for future missions to the Moon, Mars and other deep-space destinations. ### Which chip did NASA actually unveil this week? (jpl.nasa.gov) NASA identified the device as the High Performance Spaceflight Computing processor, or HPSC, in a May 12 article published by JPL and NASA’s technology site. The agency described it as a radiation-hardened, high-performance processor built as a system-on-chip for use in harsh space environments. The HPSC program is not a brand-new award. NASA’s Jet Propulsion Laboratory selected Microchip Technology of Chandler, Arizona, in August 2022 to develop the processor under a $50 million firm-fixed-price contract, with Microchip also contributing its own research and development spending, NASA said at the time. (nasa.gov) ### Is the “500 times” claim in NASA’s materials? NASA’s current public materials reviewed for this story describe the processor as delivering “up to 100 times” the computational capacity of current spaceflight computers or “over 100 times” the computing capability of current space processors. Those figures appear in NASA’s May 2026 project pages, JPL’s May 12 testing update and NASA’s 2024 HPSC white paper. (nasa.gov) NASA’s May 12 article called the processor the “centerpiece” of the HPSC project and said JPL has been subjecting it to radiation, thermal and shock testing while also running a functional test campaign. Jim Butler, HPSC project manager at JPL, said the team was “putting these new chips through the wringer” to verify they can survive spaceflight conditions. (nasa.gov) ### Why does NASA want more computing power on the spacecraft itself? NASA says communication delays beyond Earth orbit make it harder for ground controllers to handle every decision in real time. The agency’s HPSC project page says those delays drive the need for onboard autonomy for tasks involving AI, machine learning, image and signal processing, data-flow management, object detection and classification. (jpl.nasa.gov) The 2024 HPSC white paper says future missions will generate more sensor and instrument data than can be efficiently sent back for Earth-based processing, while Deep Space Network bandwidth remains limited. NASA says that combination increases the value of processing data at the edge — on the spacecraft — rather than waiting for commands and analysis from Earth. (jpl.nasa.gov) ### What would the processor let a mission do? NASA says the chip’s architecture combines computing and networking in one device and is built to support real-time autonomous tasks. On its HPSC overview page, the agency says the technology could let spacecraft process large amounts of data onboard and make decisions in real time, including driving rovers at higher speeds or filtering scientific images before transmission. (nasa.gov) Eugene Schwanbeck, a program element manager in NASA’s Game Changing Development program, said in the May 12 JPL article that the multicore system is “fault-tolerant, flexible, and extremely high-performing.” NASA’s white paper says the processor was designed for “autonomy and AI at the edge of space” and for the fault-tolerance demands of human exploration missions. (nasa.gov) ### What makes it different from older space computers? NASA says legacy radiation-hardened processors have been reliable for orbiters, capsules, telescopes and Mars rovers, but they were built on much older computing technology. The agency’s technical papers say current rad-hard processors trail commercial embedded chips by several generations, limiting their ability to run modern AI frameworks and other advanced workloads. (nasa.gov) The HPSC white paper says the new processor is a fault-tolerant, rad-hard-by-design 64-bit multicore chip with a built-in 240 gigabits-per-second time-sensitive networking Ethernet switch. NASA says its scalable design allows functions to power down when not in use, a feature meant to conserve energy on missions with tight power budgets. (jpl.nasa.gov) ### What happens next? JPL said on May 12 that testing is continuing, including radiation, thermal, shock and performance evaluations using high-fidelity landing scenarios drawn from real NASA missions. NASA has said since the 2022 contract award that the processor is being developed with the goal of use on future lunar and planetary exploration missions, and its current project pages cite missions to the Moon, Mars and beyond as target applications. (nasa.gov) (jpl.nasa.gov) (nasa.gov)