HPE Spaceborne Computer-4 Targets First HPC Benchmark on the Moon

HPE claims its Spaceborne Computer-4 will run the first HPC (high-performance computing) benchmark on the lunar surface. The compute module rides aboard Astrolab's FLIP rover, landing near the Moon's South Pole under NASA's Commercial Lunar Payload Services initiative. The claim rests on one condition: the hardware must survive the radiation and power on.

The mission launches in late 2026 on Voyager Technologies' Griffin-1 lander. Once deployed, the system will run LLM retrieval-augmented generation (AI that fetches documents to ground model answers), vision inference, LiDAR processing for mapping and navigation, and an AI services container meant to show scalable AI deployment at the lunar edge.

What hardware is actually flying?

Spaceborne Computer-4 is a smaller, lighter system than the Spaceborne Computer-1 HPE flew years ago. Mass governs every lunar payload decision, so the design targets low mass, low power, and local decision-making without round trips to Earth.

• Fanless cooling via a space radiator; the Moon has no air to move across a heatsink.
• A Kioxia BG-series M.2 SSD for storage — the same class of DRAM-less drive found in a budget TinyMiniMicro PC.
• One low-power CPU and one GPU, in a lightweight self-contained housing with an intricate lattice structure.

Kioxia has spent years proving commercial NAND on the International Space Station. This mission extends that proof to a far harsher radiation environment. That is the real test, and HPE has not published failure-rate data for the drive under lunar radiation conditions.

HPE also built software to coordinate with Earth-based resources and return results securely from the Moon. From Hawthorne, California, the company will support Astrolab's mission control using HPE Private Cloud and HPE CloudOps Software to manage and automate workloads remotely.

Who else is on board?

NASA and Deloitte will deploy additional AI workloads on the system. NASA's contribution is twofold: the FLUID experiment (Federated Learning for Unbounded and Intelligent Decentralization), which coordinates three AI models trained separately on Earth, Moon, and ISS data, plus a tuned LLM test with semantic processing to search and interpret onboard documents.

Deloitte's Silent Shield system will provide near real-time predictive cyber-analytics and anomaly detection. Whether anomaly detection proves useful or merely busy depends on how the commercial hardware degrades over the mission.

Why the benchmark matters

If the benchmarks work, they will show whether off-the-shelf-class components can support an emerging lunar economy, or whether operators must keep paying for custom space-grade silicon. That question carries real cost weight, since radiation-hardened silicon commands prices far above commodity parts.

HPE frames this as the program's step beyond the ISS into an atmosphere-free environment with far higher radiation exposure. The company calls it a milestone; the data will decide. Given the cost of lunar launches, thorough pre-flight testing is standard practice, so the primary risk sits in the environment itself, not the integration.

The wager is straightforward. Commercial NAND, a commodity SSD, and a low-power GPU stack must compute reliably where no vendor has benchmarked HPC before. If they do, the lunar economy gets a cheaper compute path; if they do not, HPE will have documented exactly where commercial silicon stops working.