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

India is poised to extend its ambitions for autonomous artificial intelligence infrastructure into space, with Chennai-based startup AgniKul Cosmos emerging as a pivotal player in domestic orbital launch capabilities. Moin SPM, co-founder and COO of AgniKul, emphasized in a recent interview that sovereignty in the AI era transcends mere data localization. "It also involves controlling access to orbital infrastructure," he stated. Moin stressed that for India to achieve genuine independence in AI capabilities, it must rely less on foreign launch services for vital missions. This gap is where private launch companies like AgniKul are stepping in. The startup, along with AI cloud provider NeevCloud, is preparing to deploy orbital AI data-centre modules in Low Earth Orbit (LEO) using an innovative design that repurposes a rocket's upper stage into a satellite bus capable of hosting various computing payloads. AgniKul points out that consistent, affordable access to orbit is crucial for developing space-based computing infrastructure. "The core of any space-oriented AI infrastructure is reliable, cost-effective access to orbit," Moin explained, noting that the advent of reusable launch vehicles could significantly alter the economics of deploying extensive satellite constellations. To facilitate this, AgniKul is designing rockets for quick turnarounds and reduced per-mission costs, which are essential for establishing the large fleets of satellites necessary for ongoing orbital computation. Their collaboration with NeevCloud aims to utilize AI server modules mounted on the upper stage of the Agnibaan rocket, effectively eliminating the need for customers to invest in specialized satellites. However, Moin cautioned that creating a scalable orbital AI infrastructure requires more than just launch capabilities. It necessitates a comprehensive ecosystem that integrates manufacturing, deployment, and operational aspects. "Constructing a robust orbital AI data backbone is a complex systems challenge," he noted, highlighting AgniKul's vertically integrated manufacturing capabilities that allow for rapid component production. Through their partnership, AgniKul and NeevCloud aim to deploy modular Space Data Centre Modules (SDCMs) in orbit, with plans to establish over 600 orbital edge data centres within three years, contingent on the success of initial missions. This initiative redefines AI sovereignty as an issue of infrastructure ownership across various domains, including land, cloud, and space. Moin outlined that a domestic launch ecosystem could support a broader private sector value chain encompassing satellite manufacturers, ground station operators, AI service providers, and data management companies. This interconnected framework is seen as vital for fostering a resilient and self-sufficient AI data ecosystem in India. The potential of space-based computing is significant, as it can process data closer to its source—satellites—thereby reducing latency and enabling real-time applications, from disaster response to enterprise-level analytics. Moin remarked, "In orbit, AI systems can process data in real-time, generating actionable intelligence within minutes of an event instead of hours, particularly in emergencies like natural disasters." India's push for space-based computing comes amidst a global race, with numerous companies, including SpaceX, Amazon, Google, and others, exploring similar orbital data-centre concepts. While ambitious satellite mega-constellations are being proposed, experts warn of the considerable technical, financial, and regulatory hurdles that lie ahead. Moin believes that establishing an indigenous launch capability could spur a comprehensive domestic value chain, resembling the way terrestrial infrastructure propelled India's IT sector. He concluded, "A credible indigenous launch ecosystem lays the groundwork for an entire private sector value chain, crucial for developing a self-reliant AI data ecosystem in India."