In 2025, NVIDIA, a global leader in graphics processing units (GPUs), sent its high-performance H100 GPU to space for the first time to evaluate the potential of operating data centers in orbit. Equipped with 80GB of high-speed memory, the GPU offers computational power that is one hundred times greater than any computer previously deployed in space. This marks a significant milestone in advancing artificial intelligence (AI) processing capabilities beyond Earth.The test mission is aboard the Starcloud-1 satellite, developed by Starcloud, a startup based in Redmond, Virginia, USA. The satellite will be launched into orbit via SpaceX’s Bandwagon 4 Falcon 9 rocket. The mission is planned to last three years, during which the satellite will operate in a low Earth orbit at an altitude of approximately 350 kilometers. While in orbit, Starcloud-1 will receive data from a constellation of synthetic aperture radar (SAR) Earth observation satellites operated by the U.S.-based company Capella, process the data in real time, and transmit the analytical insights back to Earth.

Pioneering AI Data Processing in Orbit

“The H100 is about 100 times more powerful than any GPU computer that has been on orbit before,” said Philip Johnston, CEO and co-founder of Starcloud, in an interview with IEEE Spectrum. “It will be the first time that a terrestrial-grade data-center GPU will be flown and operated in orbit.”Unlike previous deployments of NVIDIA’s Jetson machine-learning computing boards on various experimental and Earth-observing small satellites, the H100’s orbital test ushers in a new era of AI data processing in space and lays the groundwork for commercial services that could begin as early as next year.One of the key objectives of the Starcloud-1 mission is to address the challenges associated with downlinking SAR data, which is known for its massive volume. Traditionally, transmitting such data to Earth is costly and inefficient. By processing the data directly in orbit, only essential insights—such as the location, speed, and direction of a vessel—need to be sent back. This reduces the data load from hundreds of gigabytes to as little as 1 kilobyte, significantly improving efficiency and reducing communication strain.

“Downlinking SAR data has historically been a huge problem because it’s extremely voluminous,” Johnston said. “But being able to process it in orbit will mean that we can only downlink the insight. The insight might be that there is a vessel at a certain location moving at a certain speed and in a certain direction. That will be just a 1-kilobyte packet versus the hundreds of gigabytes you would need to downlink the data.”

The Case for Orbital Data Centers

Starcloud’s vision extends beyond simply processing Earth observation satellite data in orbit. The company believes that with advancements in rocket technology—particularly the anticipated cost reductions brought by SpaceX’s Starship—large-scale computing infrastructure could eventually be located in space rather than occupying valuable land resources on Earth.“As the energy demands of AI development increase, orbital data centers represent a transformative environmental breakthrough—cutting greenhouse gas emissions by orders of magnitude and eliminating the need for advanced cooling,” said Josh Parker, Head of Sustainability at NVIDIA, in an email to IEEE Spectrum.He explained that by leveraging low-cost, uninterrupted solar energy and avoiding the use of land and fossil fuels, Starcloud’s technology enables data centers to scale rapidly and sustainably. This supports the growth of digital infrastructure while helping to protect the Earth’s climate and vital natural resources.According to the International Energy Agency (IEA), global data processing infrastructure is projected to consume as much electricity by 2030 as the entire country of Japan. Meanwhile, data centers also require vast amounts of water for cooling. According to data from the World Economic Forum, a single 1-megawatt data center consumes as much water per day as roughly 1,000 people in developed countries. As AI advances drive increasing demand for computation, so too does the resource consumption of data centers. These impacts are particularly acute for communities near such facilities, which are growing increasingly concerned about rising costs and disruptions to power and water supplies. Transferring data centers to space could offer a solution to these challenges, proponents of the concept argue.

“My expectation is that within 10 years, almost all new data centers will be built in space purely because of the constraint that we’re facing on energy terrestrially,” Johnston said.

He added that achieving fully green energy operation for data centers on Earth requires significant investment in solar generation and battery storage systems. In space, however, no battery storage is necessary due to the constant availability of sunlight. Moreover, each solar panel in space generates eight times more electricity than its equivalent on Earth, further reducing costs.

“The only additional cost we have in space is the launch,” Johnston said. “We see a breakeven with launch cost of around [US] $500 per kilo. With Starship, we’re expecting launch costs much lower.”

Currently, the estimated per-kilogram launch cost for SpaceX’s Starship, once fully operational, ranges from 150toaslowas10. The rocket has successfully completed six missions to date and is expected to deploy its first batch of satellites into orbit next year.

Future Plans: Building More Powerful Orbital Computing Platforms

Starcloud is already planning its next mission, aiming to launch a data center ten times more powerful than Starcloud-1 into space next year. The Starcloud-2 mission will be equipped with NVIDIA’s upcoming Blackwell GPU as well as multiple H100 GPUs, delivering up to 7 kilowatts of computing power. Johnston said the mission is expected to provide commercial services to clients, including Earth observation satellite operators and the U.S. Department of Defense.By 2027, the company plans to launch an even larger satellite with 100 kilowatts of power. Looking ahead to the early 2030s, Starcloud envisions deploying a 40-megawatt data center in space, offering data processing at a cost comparable to that of terrestrial data centers.Starcloud is not alone in pursuing the concept of off-Earth computing. Earlier this year, Axiom Space announced similar plans. Meanwhile, Florida-based Lonestar Holdings sent a small data center to the Moon as part of the Intuitive Machines-2 mission and has ambitions to construct larger data facilities on the Moon in the coming years.