Funding & VC

India Weighs Rs 20,000 Crore Anchor Fund for Frontier AI Buildout

New Delhi is weighing a Rs 15,000-20,000 crore anchor investment in a National Frontier AI & Compute Fund to back GPU clusters, data centres and frontier model builders.

By Nathan Brooks

3 min read

Updated

Govt Eyes Rs 20kcr Fund to Power Up Frontier AI and Compute Ecosystem - tradingview.com
Govt Eyes Rs 20kcr Fund to Power Up Frontier AI and Compute Ecosystem - tradingview.comelycefeliz / Openverse

What's News

  • Government is exploring a Rs 15,000-20,000 crore anchor investment for the proposed National Frontier AI & Compute Fund under the IndiaAI Mission.
  • Epoch AI research projects training large frontier models could cost more than $1 billion per run by 2027.
  • The proposal was discussed at a closed-door Delhi meeting attended by Sarvam AI, Peak XV, E2E Networks, MeitY secretary S Krishnan and IndiaAI Mission officials.

India's government is exploring an anchor investment of Rs 15,000-20,000 crore for a proposed fund meant to give long-term risk capital to the country's frontier artificial intelligence companies and to finance GPU clusters, specialised data centres and other AI infrastructure.

The proposed National Frontier AI & Compute Fund (NFAICF), to sit under the IndiaAI Mission, remains under consultation, according to people aware of the discussions cited by the Economic Times. Its final size, structure, governance and the role of government capital have yet to be decided.

The fund is designed to pull private capital into frontier AI. The reasoning is straightforward: the cost of building models and computing infrastructure has moved rapidly beyond the reach of India's existing venture capital funds and grant-based financing.

A study by research institute Epoch AI puts the price tag for training large frontier models at more than $1 billion per run by 2027. That number frames the scale of the problem New Delhi is trying to address.

"The intent appears to be to explore a 50:50 model with private participation," said K Bhaskar Varma, chief business officer at NxtGen Datacenter & Cloud Technologies, pointing to the costs involved in building large-scale AI infrastructure.

The IndiaAI Mission already offers access to subsidised compute and funding for indigenous foundation models through its Rs 10,372-crore outlay. But the government has acknowledged a gap: there is currently no dedicated domestic equity vehicle to support frontier AI development.

Officials put the proposed NFAICF on the table at a closed-door meeting titled 'Financing India's Frontier & Compute Ecosystem', held in Delhi on Thursday.

The guest list signals the breadth of the consultation. Attendees included Vivek Raghavan, cofounder of Sarvam AI; Rajan Anandan, managing director of Peak XV; Tarun Dua, chief executive of E2E Networks; S Krishnan, secretary of the Ministry of Electronics and Information Technology; and officials from the IndiaAI Mission.

The meeting's primary purpose was to gather views from venture capital firms, infrastructure providers and other stakeholders on what role the government could play as a catalyst for frontier AI, Varma said. "They were exploring what kind of financing models could be put in place to give an impetus to the AI sector," he told ET.

Discussions also examined how private capital could be drawn into the proposed fund, and whether international institutions such as the International Finance Corporation could take part, Varma added.

Industry participants are watching the pace of execution as closely as the design. "Initiatives of this kind are essential to India's AI ambitions and should move swiftly from consultation to implementation," said Anuj Bairathi, chief executive of Cyfuture, one of the 14 cloud service providers empanelled under the IndiaAI Mission.

If the fund is finalised with the anchor investment under discussion, it would mark India's first dedicated equity vehicle for frontier AI — and a test of whether government capital can crowd in private money fast enough to keep Indian developers in a race where a single training run may soon cost a billion dollars.

Original: s3.tradingview.com

Share this article:

More from Nathan Brooks

Nathan Brooks

Show full bio

News editor covering marketplaces and e-commerce at Business Bearings.

242 articles

Related articles

« Previous articleNext article »