What is India’s first orbital data centre satellite?
The story so far:
HEOn May 4, Pixxel, a Bengaluru-based imaging satellite company, said it would partner with artificial intelligence firm Sarvam to launch a satellite called Pathfinder, described as India’s first ‘orbital data center’ satellite. This is expected to be a 200 kg class satellite, planned to enter orbit by the fourth quarter of 2026. This satellite will carry Pixxel’s hyperspectral imaging camera, which is the company’s core business, as well as data center-grade GPUs (graphics processing units).
What is an orbital data center?
It is a constellation of satellites carrying the same type of GPUs found in terrestrial data centers. Rather than simply transmitting data to ground stations, it can train and run AI models in orbit. Such a hub could perform more demanding tasks than the low-power “edge” processors that traditional satellites use for tasks such as signal compression. Edge computing on Earth means running computation close to where data is generated rather than in a central cloud, and the same logic applied in orbit is what space-based computing promises to extend.
Pixxel’s Pathfinder is being built as a single-satellite demonstrator designed to test whether ground-class hardware can operate reliably in the harsh, hot environment of low Earth orbit. “Obviously it will start as a single satellite and we will try to launch it before the end of this year,” said Awais Ahmed, the company’s CEO. Hindu.
Why are global companies suddenly interested?
Three factors have come together over the past two years and have led major technology companies to make such hubs a reality. Data centers are constrained by limitations regarding energy availability, land, water, and local regulations, all of which are amplified by the demands of artificial intelligence. In the right orbit, solar energy is effectively continuous and offers free electricity; Proponents see this as the strongest argument for moving computing into space.
Earth observation satellites also produce detailed, heavy image files that are expensive to download; Processing data in orbit and beaming down only the results has long been seen as a way to alleviate this bottleneck.
The third factor is competitive positioning. SpaceX CEO Elon Musk said that “scaling up Starlink V3 satellites with high-speed laser links on the X in 2025 will pay off. SpaceX will do that.” He also argued that “If we can solve the other parts of the equation, Starship (the company’s most powerful rocket) could deliver 100 GW per year to high Earth orbit in four to five years.” Amazon founder Jeff Bezos’ Blue Origin, Microsoft’s Azure Space and Lonestar Data Holdings have already begun pilot deployments. None of these efforts have yet created a commercial-scale orbital data center.
What are the challenges?
GPU chips run on electricity from solar panels and get hot. Now space may be cold, and common sense would suggest that it is a natural sink for heat. But space is also empty, and its emptiness eliminates convection. This is the mechanism that normally allows warm air on Earth to be transported away from terrestrial servers; A hot GPU chip in orbit is a furnace with no air to support it, unable to remove its own waste energy. The only solution to this is radiation, which requires heat to be pumped through ammonia-filled loops into deployable panels, where it can be radiated into space as infrared light. The history of crewed spaceflight is full of reminders of how unforgivable this regime can be.
Radiation damage is the second problem, and it is the one that has shaped the design of every long-duration mission ever flown. ‘Bit flips’ and long-term semiconductor degradation, where computers’ bits and bytes change randomly, are caused by cosmic rays and the radiation-hardened chips that drive most space hardware, often lagging commercial GPUs by years. Power requires storage during retention periods, and maintenance is virtually impossible without robotic service, so redundancy must be designed from the beginning.
What does the Pixxel-Sarvam partnership actually entail?
The Pathfinder satellite will be designed, built, launched and operated by Pixxel. Sarvam, an Indian AI firm, will provide what it describes as the AI backbone, with full-stack language models run on the satellite’s GPU layer for both training and inference. Pixxel’s hyperspectral camera will be carried on the same platform, giving the mission an immediate use case: Images taken in orbit can be analyzed in orbit, with only the results transmitted to Earth. Mr Ahmed declined to disclose costs, number of GPUs or launch provider, saying the choice between ISRO and SpaceX would be determined by slot availability. However, the Pixxel team includes many experts who have worked with the Indian Space Research Organization and have experience in thermal management in space.
Could processing data in space be cheaper than on the ground?
Based on current evidence, not yet and won’t be for a while. A single satellite carrying a given number of GPUs costs more than the same hardware on Earth, Mr. Ahmed said. The ultimate parity argument is based on three assumptions: that constellations will scale to tens of thousands of satellites; Once SpaceX’s Starship becomes operational, launch costs will drop sharply; and the lack of on-orbit cooling and grid power expenses will ultimately offset the higher capital expenditures. Mr. Ahmed set the horizon as 5-10 years. “It takes around 100-500 satellites to replace a data center in India and if someone pays for it, we can launch them even within 24 months,” he said. Independent reviews have been much more cautious about the timelines offered by Pixxel and its peers. Academic and agency reviews conclude that edge processing on satellites is feasible in the near term, but wholesale replacement of the terrestrial cloud is considered a 10- to 30-year proposition.
It was published – 10 May 2026 09:25 IST
