IndiaAI Mission's Compute Pillar: Democratising Access to AI Infrastructure

Core Context

The Union Cabinet approved the IndiaAI Mission in March 2024 with a total budgetary outlay of about ₹10,372 crore over five years, placing it under the Ministry of Electronics and Information Technology (MeitY). The Mission was conceived as a comprehensive ecosystem-building exercise rather than a single scheme, and it rests on seven pillars: IndiaAI Compute, IndiaAI Innovation Centre (Foundation Models), AIKosh (a datasets platform), the IndiaAI Application Development Initiative, FutureSkills, Startup Financing, and Safe and Trusted AI. The Compute pillar in particular addresses a structural constraint unique to AI development: training large models requires thousands of Graphics Processing Units (GPUs) running for extended periods, an expense well beyond the reach of most Indian startups, academic institutions and individual researchers. Rather than have the government own this hardware directly, MeitY chose a public-private empanelment model, under which private players who already own high-end GPUs are empanelled to supply computing capacity, while the government subsidises the per-hour rate charged to eligible end users.

Latest Developments

In a statement to the Lok Sabha in August 2026, the government disclosed that the Mission's shared computing capacity has now crossed 45,000 GPUs, sourced through empanelled private partners. As of August 2026, 237 AI projects spanning startups, academia and research institutions have accessed this subsidised compute, together consuming about 93.18 lakh GPU hours. Alongside this, the Mission reported that it is backing 20 indigenous sovereign AI foundation models being built by Indian teams, and has extended support to a set of Responsible AI projects under the Safe and Trusted AI pillar. A significant infrastructure milestone was also flagged: the government has placed a purchase order for a 1.1 exaflop (EFLOPS) high-performance AI computing system, to be installed at the National Informatics Centre (NIC) data centre in Delhi, which will substantially expand India's sovereign compute capacity once operational. Taken together, these numbers show the Mission moving from a policy announcement in 2024 to a functioning, measurably-utilised infrastructure programme by 2026, which is exactly the kind of implementation milestone that examiners like to convert into a factual question.

UPSC Prelims Angle

• The IndiaAI Mission is implemented by the Ministry of Electronics and Information Technology (MeitY), not by a standalone new ministry, which is a common point of confusion in MCQs on government AI initiatives.
• The Mission works through a public-private empanelment model for compute — the government does not directly purchase and own most of the GPUs; private empanelled providers own the hardware and the government subsidises access, a distinction examiners could test through an assertion-reason or statement-based question.
• AIKosh, one of the seven pillars, is specifically a data-sharing platform meant to give startups and researchers access to quality datasets — students should not confuse it with the Compute pillar.
• The 1.1 EFLOPS system ordered for the NIC Delhi data centre is a concrete, nameable fact examiners could use to test awareness of India's supercomputing and AI infrastructure landmarks.
• The Mission's Safe and Trusted AI pillar is India's institutional response to global AI governance and responsible-AI concerns, linking this topic to broader GS discussions on AI ethics and regulation.