Agnikul Cosmos Commits $60M to Build Orbital AI Data Centers in India

As the AI industry looks beyond terrestrial power grids, Indian launch startup Agnikul Cosmos is investing $60 million to manufacture orbital data centers that turn discarded rocket stages into space-based GPU clusters.
As terrestrial power grids struggle to support the massive energy demands of artificial intelligence, the space industry is racing to move compute infrastructure into orbit. Indian launch startup Agnikul Cosmos has signed an agreement to invest ₹510 crore (approximately $60 million) in a new facility in Andhra Pradesh dedicated to manufacturing orbital data centers. The move signals a major pivot for the Chennai-based company, expanding its focus from launching payloads to building the infrastructure that hosts them in space.
The new Tirupati facility will integrate GPU hardware directly into Agnikul's rocket upper stages. Instead of discarding the upper stage after deploying a satellite, the company plans to adapt it into a persistent orbital platform capable of running AI inference workloads. Bengaluru-based cloud provider NeevCloud has already signed on as the first customer, planning to supply an AI data-center module for a demonstration mission.
This approach contrasts with traditional satellite deployments by treating the rocket itself as the final product housing the compute stack. By moving AI inference into low Earth orbit, companies can theoretically bypass terrestrial energy bottlenecks and utilize continuous solar power in space. If successful, Agnikul's orbital data centers could provide a blueprint for how the commercial space sector monetizes the AI boom beyond Earth's atmosphere.
INTELLIGENCE BRIEF
WHY IT MATTERS
Moving data centers to space could alleviate the severe power constraints facing terrestrial AI infrastructure while creating a new, high-margin revenue stream for commercial launch providers.
WHO IS INVOLVED
Agnikul Cosmos, CEO Srinath Ravichandran, and cloud partner NeevCloud.
MARKET IMPACT
This bridges the gap between the commercial space sector and the AI hardware market, potentially sparking a new infrastructure race to deploy space-based GPU clusters.
This story was drafted with AI assistance and reviewed by TurkSpark editors before publication. Facts, figures, and names may be inaccurate — verify important details independently.


