HPE's Spaceborne Computer 4 to Run AI on the Moon Aboard Astrolab Rover

Fourth-gen space computer will process data at the lunar edge, perform LLM and HPC tasks on NASA CLPS mission later this year

By LineZotpaper
Published
Read Time3 min
Hewlett Packard Enterprise is sending its fourth Spaceborne Computer to the Moon later this year, aboard Astrolab's Flex Lunar Innovation Platform (FLIP) rover. The system will run AI and high-performance computing experiments near the lunar south pole, as part of NASA's Commercial Lunar Payload Services program.

Hewlett Packard Enterprise will deploy its Spaceborne Computer 4 (SBC4) on the Moon by the end of 2026, riding aboard Astrolab's FLIP rover to the lunar south pole. The rover will be delivered by Voyager Technologies' Griffin-1 lander under NASA's CLPS program.

SBC4 is a compact system measuring roughly 5x10 inches and about an inch thick, powered by an undisclosed low-power CPU and GPU. The secrecy around the chip vendor reflects the high-risk nature of the mission: if the hardware fails, the partner does not want its name tied to it, HPE told The Register. The computer will carry a pair of mirrored Kioxia SSDs totaling 1 TB of capacity.

Once on the surface, SBC4 will run a series of AI experiments including a retrieval augmented generation (RAG) system to search onboard documents, and machine learning algorithms processing camera and LiDAR data. NASA will also put the system through high-performance computing benchmarks: high performance Linpack (HPL) and high performance conjugate gradient (HPCG). HPE and Astrolab expect this to be the first Top500 submission made from a celestial body.

"It's important to ship back insight to Earth rather than all the raw data whenever possible," said Astrolab CEO Jaret Matthews. "Having the edge compute capability to process the raw data locally, save on bandwidth back to Earth."

HPE's chief scientist of space tech and solutions, Mark Fernandez, said the system uses "hardware hardening in software" to cope with radiation and temperature swings. Off-the-shelf components are protected by a software RAID 1 mirror and continuous monitoring against a state table. "I like to be running at full speed. I can't do that. I'll run at a slower speed," he explained, describing the system's self-preservation steps. HPE drew on nine years of experience with Spaceborne Computers on the International Space Station.

The rover will also benefit from the heat generated by HPL workloads during the Moon's 14-day night cycles, potentially preventing icing.

Bandwidth constraints make edge computing essential: the round-trip latency to the Moon is about 2.5 seconds, and high-resolution imagery cannot be transmitted in real time. SBC4 is designed "at the edge of the edge," Fernandez said.

While companies like Google, Amazon and SpaceX have touted orbital datacenters, the economics of getting such facilities into low Earth orbit remain unproven, The Register notes. HPE's approach focuses on practical edge compute for specific missions.

§

Analysis

Why This Matters

  • This is the first time AI and supercomputing benchmarks will run on the lunar surface, demonstrating how space missions can process data locally rather than relying on slow Earth links.
  • Success could pave the way for more autonomous rovers and landers, reducing dependence on ground control for time-sensitive decisions.
  • The mission tests whether commercial off-the-shelf components can survive the Moon's radiation and temperature extremes, potentially lowering costs for future lunar infrastructure.

Background

Hewlett Packard Enterprise has been building space-rated computers for the International Space Station since 2017, with its Spaceborne series gradually increasing capabilities. The FLIP rover is Astrolab's first lunar vehicle, developed under NASA's Commercial Lunar Payload Services program, which contracts private companies to deliver payloads to the Moon. The south pole region is of interest for water ice and as a candidate for future crewed bases. SBC4 is one of several technology demonstrations on FLIP.

Key Perspectives

HPE and Astrolab: They argue that edge computing is essential for lunar operations given bandwidth and latency constraints. Processing data locally, rather than sending raw imagery back to Earth, saves time and allows real-time analysis. HPE believes off-the-shelf hardware, paired with clever software redundancy, can withstand the lunar environment.

NASA: The agency gains a test bed for AI and high-performance computing in a realistic off-world setting. The LLM and RAG experiments could inform how astronauts and ground crews interact with documentation on future missions.

Skeptics: The secrecy around SBC4's chip vendor suggests reliability concerns. Earlier Spaceborne Computers suffered SSD failures caused by bad capacitors, not space radiation, and while HPE says it has had no failures since 2019, the Moon's environment is harsher than low Earth orbit. The economics of space-based compute remain unproven; Google, Amazon and SpaceX have proposed orbital datacenters, but such projects have not materialised.

What to Watch

  • Whether the chip vendor is revealed once SBC4 is operational on the Moon.
  • The results of the HPL and HPCG benchmarks: they will provide the first data on supercomputer-class performance beyond Earth.
  • How well the software-based radiation hardening holds up over the year-long mission, especially during solar flares or the two-week lunar nights.

Sources

Checked

12 claims in this story checked against 1 source: 5 supported.

Show each claim
  1. SupportedSBC4 is slated to run AI experiments on the Moon aboard Astrolab’s FLIP rover.www.theregister.com - Articles: “Assuming a safe landing on the surface, FLIP will serve as a test bed for a range of emerging technologies including HPE’s latest space-faring compute platform.”
  2. SupportedSBC4 will perform LLM and HPC tasks on a NASA CLPS mission later this year.www.theregister.com - Articles: “Spaceborne Computer 4, which for the sake of brevity we’ll just call SBC4, will carry out a number of AI and HPC-centric experiments during the rover’s year long mission on the lunar pole.”
  3. Partly supportedMatthews said, “Having the edge compute capability to process the raw data locally, save on bandwidth back to Earth.”www.theregister.com - Articles: ““It’s important to ship back insight to Earth rather than all the raw data whenever possible. So having the edge compute capability to process the raw data locally, save on bandwidth back to Earth,” Astrolab CEO Jaret Matthews said.”
  4. Partly supportedHPE told The Register that the chip-vendor secrecy reflects mission risk and the partner’s reluctance to be named if the hardware fails.www.theregister.com - Articles: “We’re told the reason for the secrecy largely comes down to the high-risk nature of the mission. If the hardware fails, HPE’s compute partner doesn’t want their name tied to it.”
  5. Reports differVoyager Technologies’ Griffin-1 lander will deliver the rover.www.theregister.com - Articles: “Part of NASA’s Commercial Lunar Payload Services (CLPS) program, the rover is set to be deployed near the Moon’s south pole by the Voyager Technologies Griffin-1 lander later this year.”
  6. Partly supportedSBC4 will be on the Moon by the end of 2026.www.theregister.com - Articles: “Before the end of the year, Hewlett Packard Enterprise (HPE) will have AI running on the Moon.”
  7. Partly supportedSBC4 uses an undisclosed low-power CPU and GPU.www.theregister.com - Articles: “What we do know is that it is equipped with a low-power central processing unit, a graphics processor”
  8. SupportedSBC4 will use mirrored Kioxia SSDs with 1 TB of total capacity.www.theregister.com - Articles: “it is equipped with a low-power central processing unit, a graphics processor, and will be equipped with a pair of mirrored Kioxia SSDs totaling 1 TB of capacity.”
  9. Partly supportedHPE drew on nine years of experience with Spaceborne Computers on the ISS.www.theregister.com - Articles: ““With our nine years' experience in the Space Station using drives, we’ve even convinced them we don’t need to do anything special for space.””
  10. Partly supportedThe rover will benefit from heat generated by HPL workloads during lunar night, potentially preventing icing.www.theregister.com - Articles: “a workload like HPL could have other benefits like helping to keep the rover from icing over during the roughly 14-day-long night cycles”
  11. SupportedHPE and Astrolab expect their submission to be the first Top500 submission made from a celestial body.www.theregister.com - Articles: “which means HPE and Astrolab will soon hold the distinction of the first Top500 submission made from a celestial body.”
  12. SupportedThe round-trip latency to the Moon is about 2.5 seconds.www.theregister.com - Articles: “The latency between Earth and the Moon is nowhere near as bad as Mars's up to 44 minute round-trip communication time, but even 2.5 seconds makes real-time control from Houston, or in this case Astrolab’s mission control, which by the way is also powered by HPE's Greenlake Private Cloud stack, in Hawthorne, California a bit unwieldy.”

Zotpaper

Written by Rex Doyle on the Science desk. Claims checked by Ines Fairley, verification. Edited by Angus Burke, chief sub. Plate by Ollie Flash, picture desk.

1 source5 of 12 claims matched to a sourcefirst published
Meet the newsroomHow we workSubscribe