When India’s AI Ambitions Met Fiscal Reality
Artificial intelligence dominated the government’s policy narrative this year, but the spending choices point to a far more cautious strategy focused on application and infrastructure rather than frontier models.
News
- UK Regulator Seeks Explanation From Air India Over Dreamliner Switch Issue
- Alphabet Plans Major Bengaluru Expansion With Space for 20,000 Staff
- Amazon’s AI Push Moves into Hollywood’s Production Pipeline
- Alphabet Raises AI Stakes With $185 Billion Spending Plan
- Peak XV Loses Three Senior Investors as Veteran Trio Spins Out
- DGCA Rules Out Fault in Air India Dreamliner Fuel Control Switch
Artificial intelligence loomed large in the government’s economic policy narrative this year. Yet a closer reading of the allocations points to a strategy that is more restrained than the rhetoric initially suggested.
Finance minister Nirmala Sitharaman repeatedly invoked AI—about 11 times—in this week’s Union Budget, linking it to public services, agriculture, customs checks, school education, skilling and employment platforms. The emphasis was firmly on use cases rather than breakthroughs. Unlike the US or China, the government avoided announcing large-scale investments aimed at building frontier AI models.
That caution becomes clearer in the spending plans. Allocations under the IndiaAI Mission have come in below earlier signaling. The revised estimate for the current fiscal that ends next month (fiscal year 2025–26) stands at ₹800 crore, well below the ₹2,000 crore initially earmarked.
For the next fiscal (FY27), the allocation rises to ₹1,000 crore, still short of expectations set last year when the Union Cabinet approved a five-year outlay of a little more than ₹10,300 crore.
The composition of spending points to a deliberate policy choice. More than 85% of the roughly ₹1,100 crore spent so far under the mission has gone towards subsidizing access to computing infrastructure for domestic developers. Compute alone now accounts for over 40% of total IndiaAI Mission spending.
Several indigenous AI firms, including Sarvam, BharatGen, Zenteq, SoketAI, Gnani.ai, Gan.ai and Avataar AI, have benefited from these subsidies. The approach reflects an attempt to treat compute as a shared national input, lowering entry barriers for local players without committing the state to backing specific models.
Other elements of the mission have moved more slowly. Funding for datasets, skilling, startups and governance frameworks remains limited, suggesting that while servers are being brought online, the broader ecosystem needed to sustain AI development is still catching up.
The government’s public stance on AI also appears to be evolving. Last year, officials spoke of delivering an Indian equivalent of China’s DeepSeek moment, where engineering efficiencies produced strong results at relatively low cost. Since then, the tone has grown more measured.
The Economic Survey cautioned against scaling AI deployments without clear economic purpose. In January, information technology minister Ashwini Vaishnaw said that many real-world needs could be met using smaller language models rather than capital-intensive, power-hungry systems.
Ankush Sabharwal, founder and chief executive of CoRover, says that the lower headline allocation does not signal a retreat. Of the ₹2,000 crore set aside last year, only about ₹800 crore was actually spent, he notes. Seen this way, the ₹1,000 crore now allocated represents an increase on real outlays.
“The Union Cabinet approved over ₹10,300 crore for the IndiaAI Mission in March 2024 to be spent over five years,” Sabharwal said. “This year’s ₹1,000 crore allocation is actually 25% higher than last year’s real spending, since only about ₹800 crore of the ₹2,000 crore earmarked was used.”
Others remain skeptical. Rohit Kumar, founder of Delhi-based public policy consulting firm The Quantum Hub, said the program appears to have lost momentum at a critical stage. Earlier talk of expanding the mission to ₹20,000 crore has given way to tighter allocations, he said, risking slower progress when domestic AI capability still requires sustained support.
At the same time, policy attention appears to be tilting toward physical infrastructure. Between October and January, companies announced more than $75 billion in data center investments, with tax incentives for such facilities featuring prominently in the fiscal plan.
The implicit bet is that infrastructure and electronics manufacturing will deliver faster and more predictable economic returns than large public investments in AI model development.
India has not stepped away from artificial intelligence altogether. The Ministry of Electronics and Information Technology was allocated ₹21,632 crore, leaving room for digital infrastructure and applied technology programmes even as headline AI funding remains limited.
The approach points to a preference for diffusion over dominance, embedding AI across public services and industry rather than racing to build frontier systems at scale, analysts added.
