India wants AI sovereignty. Agnikul says that means going to space

India’s push for sovereign artificial intelligence infrastructure may soon extend beyond terrestrial data centres to space, with Chennai-based startup AgniKul Cosmos positioning itself as a domestic launch backbone for orbital compute.

“Sovereignty in the AI age is not just about data localisation on the ground,” Moin SPM, co-founder and chief operating officer of AgniKul Cosmos, told Business Today in an interview. “It extends to who controls access to the infrastructure in orbit.” 

He added that if India wants truly independent AI capabilities, it must avoid reliance on foreign launch providers for critical missions, a gap private launch startups are now seeking to fill. 

The comments come as AgniKul and AI cloud firm NeevCloud plan to deploy orbital AI data-centre modules in Low Earth Orbit (LEO), using a patented design that converts a rocket’s upper stage into a satellite bus capable of hosting computing payloads.

Launch capability as the foundation of orbital AI

According to AgniKul, access to orbit, frequent, reliable and affordable, is the primary bottleneck for space-based computing infrastructure.

“The foundation of any space-based AI infrastructure is reliable, affordable, and frequent access to orbit,” Moin said, adding that reusable launch vehicles fundamentally change the economics of deploying large constellations. 

The company is developing rockets designed for rapid turnaround and lower per-mission costs, which could enable the large fleets of satellites required for persistent orbital compute.

Through its collaboration with NeevCloud, Agnikul plans to host AI server modules on the extendable upper stage of its Agnibaan rocket, a design the company says eliminates the need for customers to build dedicated satellites. 

Building an orbital AI data backbone

Scaling space-based AI infrastructure requires more than launch vehicles, Moin argues. It demands an integrated ecosystem spanning manufacturing, deployment and operations.

“Building a scalable orbital AI data backbone is not a single engineering challenge. It is a systems challenge,” Moin told Business Today. 

He pointed to the company’s vertically integrated manufacturing and additive-manufacturing capabilities, which allow components to be produced in days rather than months.

NeevCloud plans to leverage these capabilities to deploy modular Space Data Centre Modules (SDCMs) in orbit. The firms aim to scale to more than 600 orbital edge data centres within three years if initial missions succeed.

From data localisation to orbital control

The concept reframes AI sovereignty as a question of infrastructure ownership across domains, land, cloud, and now space.

A domestic launch ecosystem, Moin said, could anchor a broader private-sector value chain including satellite manufacturers, ground-station operators, AI service providers, and data companies. 

“That multiplier effect is what will ultimately build a resilient and self-reliant AI data ecosystem for India,” he said. 

This argument resonates with policymakers increasingly concerned about dependence on foreign hyperscalers and satellite networks for critical digital infrastructure.

Real-time intelligence from orbit

According to the company, space-based compute could process data closer to its source (satellites), reducing latency and enabling real-time applications ranging from disaster response to global enterprise analytics.

“In orbit, AI systems can process data closer to the source, reducing latency and enabling real-time decision-making that is simply not possible when data has to travel back to ground-based infrastructure,” Moin told Business Today. 

He added that such capabilities could generate actionable intelligence “within minutes of an event, rather than hours” for emergencies such as natural disasters. 

A crowded global race for space compute

India’s effort comes amid intensifying international competition to move data processing into space. Multiple companies, including SpaceX, Amazon, Google, Blue Origin, Axiom Space and others, are exploring orbital data-centre concepts, while new satellite mega-constellations are being proposed.

Several proposals involve tens of thousands to hundreds of thousands of satellites, though experts caution that technical, financial, and regulatory challenges remain formidable.

Moin argues that indigenous launch capability could catalyse an entire domestic value chain, from satellite manufacturing to AI platforms, similar to how terrestrial infrastructure enabled India’s IT industry.

“A credible indigenous launch ecosystem creates an anchor around which an entire private sector value chain can develop,” he said, spanning satellite builders, ground stations, data service providers, and AI companies. 

He described this multiplier effect as essential to building “a resilient and self-reliant AI data ecosystem for India.”