Indian Startups Sarvam and Pixxel Plan AI-Powered Data Centre in Space
News Mania Desk/ Piyal Chatterjee/ 4th May 2026
In a significant step towards next-generation computing, Indian startups Sarvam AI and Pixxel are working together to develop an artificial intelligence-powered data centre in space, marking a pioneering move in the country’s tech and space ecosystem.
The proposed project involves launching a satellite-based “orbital data centre” that will be capable of processing data directly in space using advanced AI systems. Unlike conventional satellites that transmit raw data back to Earth for analysis, this system aims to carry out real-time processing in orbit, significantly reducing delays and improving efficiency.
Sarvam AI will provide the artificial intelligence infrastructure, including its AI models and computing stack, while Pixxel will be responsible for building, launching and operating the satellite. The collaboration is expected to combine expertise in AI and satellite technology to create a platform capable of handling complex data tasks beyond Earth.
The initiative is part of a growing global effort to move computing infrastructure into space. Traditional data centres on Earth face increasing challenges such as high energy consumption, cooling requirements and space limitations. In contrast, space offers advantages like abundant solar power and natural cooling conditions, making it an attractive alternative for energy-intensive AI operations.
The satellite will be equipped with high-performance computing systems similar to those used in terrestrial data centres, enabling it to run AI models and analyse large volumes of information in real time. This could prove particularly useful in areas such as Earth observation, climate monitoring and defence, where rapid data processing is critical.
Experts believe that in-orbit computing could transform how satellite data is used, eliminating the need to send vast amounts of information back to ground stations before analysis. This would not only save time but also reduce bandwidth constraints and operational costs.
