Nvidia's push to place its computing hardware inside more physical machines is moving beyond Earth. Lunar Outpost, a startup building robotics for space infrastructure, says its next moon rover will use Nvidia Jetson chips to control its lidar system, a step that is likely to put the first GPU on the lunar surface.
The plan is not just about sending a chip to a dramatic destination. It is a test of whether more capable GPU-powered systems can operate in an environment where radiation, temperature swings, low power and long periods of darkness shape what machines can do.
Why Jetson is going to the moon
Lunar Outpost is comparing Nvidia Jetson with its existing flight compute platform, which CEO Justin Cyrus described as having more spaceflight heritage. The company is weighing the strengths and drawbacks as it works toward using GPU-powered systems in extreme environments.
Jetson is not as widely known as Nvidia's Blackwell and Vera Rubin platforms, but it plays an important role in physical AI. The platform is designed to be compact and power-efficient, which makes it relevant for robots that must process sensor information where they are, instead of relying on distant systems to interpret the world for them.
On the rover, Jetson will control a lidar system. Lidar can help a machine understand nearby terrain, and local processing can help a robotic system respond more quickly to what its sensors detect. For a rover moving through difficult lunar terrain, that kind of onboard awareness matters.
Cyrus said Lunar Outpost's autonomy stack has changed over time: "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." He added that the company still runs both approaches in parallel as it figures out where physical AI can help.
NASA's private lunar push sets the stage
The rover is part of a wider effort to explore the moon before a possible return of human astronauts, perhaps as soon as 2028. NASA has launched a campaign to pay private companies to explore the lunar surface, using a model based on its work with SpaceX.
Under that program, tech companies are being asked to build vehicles that can carry scientific sensors to the moon. The goal is to better understand the terrain and look for substances such as water, which could be useful or potentially lucrative to exploit.
Lunar Outpost's next rover will ride aboard a lander built by Intuitive Machines, another company focused on the moon. The rover is meant to bring a package of sensors into craters and other areas that are hard to study from orbit.
A following mission is planned for Reiner Gamma, a place on the lunar surface where a magnetic anomaly has puzzled scientists. Each mission is expected to launch on a Falcon 9 rocket before the end of the year.
Nvidia has another lunar computing test
Nvidia's lunar work is not limited to the Lunar Outpost rover. The company also recently announced a partnership with Firefly Aerospace, the first private company to safely land a robot on the moon.
That partnership will put the Jetson platform on a satellite orbiting the moon, where it will process imagery. The satellite is intended to gather data for scientists working to map the moon, while also tracking the growing number of robots operating on the lunar surface.
Together, the rover and satellite efforts point to a practical question: can compact GPU systems do useful work in and around the moon without depending entirely on Earth-based computing? The answer matters because the moon is becoming a place where more robotic activity is expected, and those machines will need to understand their surroundings.
The lunar environment is the hard part
Putting GPUs in space is difficult. Many GPUs that operate in space are closer to Earth, where they receive more protection from radiation and dramatic temperature shifts.
The lunar surface is harsher. The moon is more directly exposed to cosmic radiation, and temperatures change sharply as it moves through its phases. Cyrus summarized one of the central constraints plainly: "Your system has to survive lunar night and has to do so on very low power."
That makes this effort a test of both performance and endurance. A chip that can quickly process sensor data is useful only if the wider system can keep working through the moon's environmental stress.
If Lunar Outpost can make the chips operate there, its engineers hope the company's vehicles will become more capable. Faster environmental processing could help rovers make decisions more quickly, which is especially important when robots are expected to enter places that are difficult to explore from orbit.
From exploration to a robotic workforce
The long-term ambition is larger than a single rover. NASA has broad plans for a sustained human presence on the moon, and that would likely require autonomous systems to prepare the way and handle significant work before and alongside astronauts.
Cyrus framed Lunar Outpost's work around that transition: "What we at Lunar Outpost are currently working on is how do we go from exploration to permanence, how do we actually build that outpost on the moon?" He said the combination of deterministic models and a physical AI layer could help make human presence in space sustainable through a robotic workforce.
Lunar Outpost has several smaller autonomous rovers scheduled for launch to the moon in the next few years. It also has a larger rover, Pegasus, intended to carry astronauts.
Pegasus is waiting for a rocket from Blue Origin, Jeff Bezos's space company. That launch vehicle suffered an anomaly this summer, and it is not clear when it will return to flight. Cyrus said he could not answer Blue Origin's timelines or NASA's timelines on pad reconstruction, but said Lunar Outpost has been told it remains on the same timeline for 2028.
For now, visions of space data centers and fleets of lunar robots depend on the same foundation: reliable transportation to the moon, and computing systems that can survive once they get there.