HPE computer will run first LLMs on the moon aboard Astrolab rover

A high-performance computer built by Hewlett Packard Enterprise will run large language models on the lunar surface for the first time, riding aboard Astrolab's 480-kilogram FLIP rover scheduled to launch as early as November 2026.

The mission, FLEX Lunar Innovation Platform, is the first in a series of rovers NASA has contracted from Astrolab, including future vehicles designed to carry Artemis astronauts across the lunar surface. FLIP will carry 20 scientific payloads, among them a lidar 3D mapper and a sensor measuring helium-3 concentrations in lunar regolith. The rover's deformable wheels target rugged terrain, and its systems must survive at least 150 hours of the frigid lunar night.

The HPE Spaceborne-4 edge computer pairs a 64-bit CPU with a GPU capable of running large numbers of computational tasks in parallel — architecture comparable to advanced edge computers on Earth, not the decades-old flight computers typical of planetary rovers.

Why run LLMs on the moon?

Custom-built LLMs aboard FLIP will run a series of experiments to prove the technology can operate reliably in the moon's extreme environment. The models will analyze data from onboard cameras and the lidar mapper to aid safe navigation and distill insights from gathered data, cutting the volume of information the rover must transmit home.

The business case is latency. "The time delay, for example, on Mars can be up to 20 minutes," Mark Fernandez, chief scientist and principal investigator at HPE, said during a press conference. "On the moon it's reduced, but still, you are looking at stale data. In fact, the Soviets lost two rovers in the past because of operator errors caused by the time delay."

Autonomy is not new in space. China's Chang'e-6 rover, which landed on the far side of the moon in 2024, used basic AI algorithms and neural networks for autonomous navigation. NASA's Mars rovers Curiosity and Perseverance have driven semiautonomously on pre-programmed paths for years — but their space-hardened computers are remarkably weak, comparable to late-1990s laptops.

From ISS to lunar south pole

HPE has flown three Spaceborne computers to the International Space Station. Spaceborne-2 and Spaceborne-3, launched in 2021 and 2024, ran Meta's Llama 3.2 aboard the ISS as part of the "Space Llama" initiative in 2025, helping astronauts streamline research tasks and speed decision-making.

Rather than spend years building specialized space-hardened hardware, HPE deployed software that monitors the system and detects upsets caused by charged cosmic particles. When radiation spikes, the software lowers the computer's power and speed to protect the electronics. "In space, we don't have the redundancy, we don't have the repairman, we don't have the spare parts," Fernandez said. "So we spent a lot of time, a lot of CPU cycles, checking everything out and making sure that everything is working out right."

The approach has outperformed expectations: each ISS computer completed runs of more than two years. The moon raises the bar considerably. "The ISS is an extreme environment, but it's relatively controlled in comparison to what we're going to get on the moon," said Francisco Cordova, chief operating officer for the ISS National Laboratory.

FLIP will land near the lunar south pole, where temperatures swing between 80 °C in sunlit areas and -200 °C in shaded regions the rover will search for water ice. Cosmic radiation on the lunar surface runs almost three times as intense as inside the shielded station, which also benefits from Earth's magnetic field.

Spaceborne-4 keeps the same software-protected architecture and uses the rover's radiators to shed excess heat. "Our expectation is that the computer is going to last for the life of the mission," said Scott Wiest, HPE's chief technology officer, referring to the planned 10-day endeavor.

What could go wrong?

Independent researchers urge caution on reliability claims. Roser Almenar, an AI and space law researcher at the University of Valencia, said LLMs hold "an enormous potential, in particular for future missions to Mars and beyond, where communication delays will make it increasingly difficult to rely on immediate assistance from Earth." But she noted it is early days: "These systems can produce inaccurate or misleading information, and in space, even a seemingly small error can have serious consequences."

Cordova framed the longer arc: "As we venture further and further out into space, I believe that we're going to rely more on AI for space exploration, for making smarter and better decisions, but more importantly, working in those environments where humans simply cannot."

The FLIP mission's LLM experiments — measured against a 10-day window near the lunar south pole — will supply the first operational data on whether commercial edge computing and language models can carry that weight.