Union Budget 2026: AI Industry Wants Cheaper Compute, More IndiaAI Funding, and Clearer Incentives

With Union Budget 2026 set to be announced shortly, India’s AI ecosystem is watching for signals that move beyond broad “digital” spending and directly address what startups and enterprises say is the biggest bottleneck: affordable access to compute. Industry voices have repeatedly flagged high GPU costs and limited domestic compute capacity as constraints on building and deploying large-scale AI in India.

1) Bigger outlay for IndiaAI Mission and public compute

A key expectation is a higher allocation for the IndiaAI Mission to expand national compute capacity and make it accessible to startups, researchers and public institutions. One set of pre-budget demands calls for raising the IndiaAI Mission allocation to at least ₹5,000 crore annually from around ₹2,000 crore, with a focus on scaling GPU availability.

2) A “compute credit” or subsidy model

Instead of only funding infrastructure, the industry wants a mechanism that directly lowers the cost of using it, especially for early-stage companies. Proposals include a compute credit scheme (for example, time-bound GPU hours or a subsidy on approved usage) so founders can train and fine-tune models without being priced out by cloud bills.

3) Duty and tax relief for AI hardware and data centres

Data centres and AI infrastructure are power- and capex-heavy, so fiscal easing is another major ask. Industry-facing proposals include customs duty rationalisation for imported critical equipment like GPUs, along with tax measures, such as better GST input credit treatment on data-centre capital assets.

4) Targeted skilling beyond short courses

AI skilling remains a budget-season theme, but the demand is to fund programmes that translate into deployable talent—engineers who can work on applied AI, data engineering, model operations and security. There are also calls to reduce the tax burden on AI-specific vocational training (including proposals around GST).

5) Incentives for Indian-language models and domestic datasets

Another expectation is more direct support for India-specific AI, models and datasets that work across Indian languages and public-service contexts. Industry leaders have argued that policy needs to move from intent to execution, including clearer rules around consent and “data dignity,” while incentivising applied AI adoption across sectors.

If Budget 2026 delivers even one decisive intervention, predictable, subsidised access to compute, it could set the pace for how quickly Indian AI moves from pilots to scaled products and services.

Read more: Government Reportedly Plans to Ban Social Media for Kids in Upcoming Bill