Nvidia’s efforts to widely deploy its GPUs are aiming for a new goal: the moon.
Lunar Outpost, a startup building robotics for space infrastructure, announced Thursday that its next lunar rover will use a Jetson chip to control its lidar system. If realized, this would likely be the first GPU on the moon.
“We’re taking Nvidia Jetson and comparing it to our flight computing platform, which has a bit more of a spaceflight heritage,” said Justin Cyrus, CEO of Lunar Outpost. “(We’re) looking at what the strengths are and what the weaknesses are, and we’re committed to deploying systems with higher performance GPUs in these extreme environments.”
NASA has launched a campaign to get private companies to fund lunar exploration ahead of plans to return human astronauts, possibly as early as 2028. The space agency’s program, modeled on its work with SpaceX, calls on technology companies to develop vehicles to carry scientific sensors to the moon to understand the moon’s topography and search for materials like water that could be useful or even profitably exploited.
Nvidia also recently announced a partnership with Firefly Aerospace, the first private company to safely land a robot on the moon. This will result in the Jetson platform operating on a lunar orbiter to process images. The satellite is intended not only to collect data for scientists trying to map the moon, but also to track the growing number of robots on the moon’s surface.
Lunar Outpost’s next rover will be mounted on a lander built by Intuitive Machines, another moon-focused technology company. The spacecraft is designed to carry packages of sensors to craters and other locations that are difficult to explore from orbit. The next mission will explore a place on the moon called Liner Gamma, where a magnetic anomaly has baffled scientists.
Each mission is scheduled to fly on a Falcon 9 rocket by the end of the year.
Nvidia’s Jetson platform isn’t as well-known as AI workhorses like Blackwell and Vera Rubin, but it’s a key part of many physical AI systems. Jetson is designed to be compact and power efficient, allowing robotic systems to process sensor input locally, allowing them to more quickly understand and react to the world around them.
“Our autonomy stack was a little more deterministic five years ago, but now it’s a combination of deterministic AI and physical AI, which is really fun,” Cyrus says. “We’re still running both in parallel, and it’s our job to understand where physical AI can actually plug into our stack and help us do things that no one has done before.”
Using GPUs in space is difficult due to the extreme environment. Most space-flight GPUs are in orbit close to Earth, where they are relatively protected from radiation and rapid temperature changes. The moon is directly exposed to cosmic radiation, and its temperature fluctuates depending on the moon’s phases.
“The system has to survive a lunar night, and it has to run at very low power,” Cyrus says.
If they can get the chip to work in that environment, Lunar Outpost engineers hope it will improve the vehicle’s capabilities, allowing it to process the environment and make decisions more quickly. NASA has broader ambitions to support a continued human presence on the moon, but that will require autonomous systems to lead the way and do the heavy lifting.
“What we’re working on right now with the lunar outpost is how do we move from exploration to a permanent presence and how do we actually build an outpost on the moon,” Cyrus said. “We believe that by combining deterministic models with a physical AI layer, we can make human presence in space sustainable and actually enable a robotic workforce.”
In the coming years, the lunar outpost is scheduled to launch several small autonomous rovers to the moon, and one large autonomous rover intended to carry astronauts. Pegasus is waiting to take a rocket built by Jeff Bezos’s space company Blue Origin to the moon, but its launch vehicle suffered an anomaly this summer and it’s unclear when it will return to flight.
“They seem to be making pretty good progress on the pads,” Silas said. “I can’t give you an exact answer about Blue Origin’s timeline or NASA’s timeline for pad reconstruction, but everything we’ve been told is we’re on the same timeline (of 2028).”
So we’re left with futuristic visions of space data centers and swarms of robots working on the moon, both looking forward to the same thing: bigger rockets.
If you buy through links in our articles, we may earn a small commission. This does not affect editorial independence.
