Nvidia chips are heading to the Moon on a robot rover

Key points

• Lunar Outpost, a robotics startup, announced on Thursday that its next Moon rover will use Nvidia Jetson chips to control its sensor system.
• If successful, these will be the first GPUs (the specialised chips that power AI processing) to operate on the lunar surface.
• Nvidia has separately partnered with Firefly Aerospace, the first private company to land a robot safely on the Moon, to run Jetson chips on a lunar-orbiting satellite.
• Two Lunar Outpost rover missions are each expected to launch on a SpaceX Falcon 9 rocket before the end of 2025.
• A larger crewed rover called Pegasus is targeting a 2028 Moon landing, but depends on a Blue Origin rocket that suffered a malfunction this summer.

A small AI chip could soon be sitting on the Moon's surface, millions of miles from the nearest data centre.

Lunar Outpost, a startup that builds robotic vehicles for space, announced Thursday that its next Moon rover will carry an Nvidia Jetson processor. The Jetson is a compact, low-power chip designed to let robots process sensor data on the spot, without sending it back to Earth to be analysed. On a Moon rover, that speed matters enormously.

Why send an AI chip to the Moon?

The short answer: rovers need to think for themselves. Radio signals between Earth and the Moon take more than a second each way, so a robot that has to phone home before making a decision moves dangerously slowly across rocky, crater-pocked terrain.

The Jetson chip runs what engineers call a "physical AI" layer, software that reads the rover's sensors and works out how to move through an unfamiliar environment without a human directing every step. Lunar Outpost CEO Justin Cyrus described it as combining that kind of flexible AI reasoning with older, more rigid rule-based systems that the company has trusted for years.

"Our autonomy stack was a bit more deterministic five years ago, and now it's a combination of deterministic and physical AI, which is pretty fun," Cyrus told TechCrunch AI.

What are the two missions actually doing?

The first rover rides aboard a lander built by Intuitive Machines and is built to carry scientific sensors into craters and other spots that satellites cannot easily see from orbit. The second mission heads to a place called Reiner Gamma, a patch of the lunar surface where an unusual magnetic field has puzzled scientists for decades. Both launches are scheduled on SpaceX Falcon 9 rockets before the end of this year.

Nvidia also announced a separate deal with Firefly Aerospace to put Jetson chips on a satellite orbiting the Moon, where they will process images to help scientists map the surface and track the rising number of robots working there.

Is running a GPU on the Moon actually hard?

Very. Most AI chips that operate in space sit in low Earth orbit, where the planet's magnetic field blocks much of the harmful radiation coming from the sun and deep space. The Moon has no such shield.

Temperatures on the lunar surface swing by hundreds of degrees as the Moon cycles through its roughly month-long day and night. During lunar night, a rover must survive on very little power and then wake up intact. Chips designed for warehouses and research labs were not built for any of that.

Lunar Outpost is currently testing the Jetson against its existing flight-rated hardware to see which holds up better.

Looking further ahead, the company is building a larger rover called Pegasus that is designed to carry astronauts. That vehicle is waiting on a rocket from Blue Origin, Jeff Bezos's space company, which suffered a launch-pad problem this summer. Cyrus said he is cautiously optimistic the timeline to 2028 still holds.

What to watch for: If the Jetson chips survive both rover missions intact, expect more capable AI hardware to appear on future lunar vehicles. That would be a meaningful step toward the autonomous robot workforce that NASA considers essential before humans return to the Moon.