IndiaAI’s Equity in AI Startups is a ‘Fantastic’ but Risky I…

After Sarvam AI, the government chose Soket AI Labs, Gnani AI, and Gan AI last month as part of the IndiaAI Mission to build India’s sovereign AI. As part of the agreement, the government will receive equity stakes (ranging from 2% to 4%) in each of these startups, as per reports.

This move has triggered concern across the AI and investor ecosystem, though some founders have expressed relief over finally getting capital support for high-R&D workloads.

“This is not a grant. A government body will take equity in Sarvam for the compute we receive. We are committed to building public interest use-cases and enabling the ecosystem in various ways, such as hyper-optimising the inferencing costs in India,” said Pratyush Kumar, co-founder of Sarvam AI, during the selection announcement.

While Sarvam has already released a few updates and models, the other selected startups haven’t released anything yet and are still awaiting the promised GPU support from the government. Abhishek Upperwal, founder of Soket AI Labs, told AIM that though the amount of equity is not yet confirmed, he is optimistic about the government’s role in the startup’s journey. 

“In our case, it’s actually helpful. For the kind of R&D intensive work we aim to do, VCs don’t have the appetite.” Upperwal said this would give the startup access to GPUs, not readily available through VC funding in India.

Indian investors also lack the deep pockets to back startups in training foundation models for AI, like the one Soket AI is developing, which has 120 billion parameters. But that’s not the only worry.

What’s the Issue?

Some say the “grant” should have gone to a public initiative like AI4Bharat. Vivek Raghavan, one of Sarvam’s co-founders, is also a mentor at AI4Bharat and an advisor to Bhashini.

Siddharth Sharma, former CTO of Shaadi.com, argued earlier that the government shouldn’t have picked for-profit companies at all. “The only way equity makes returns is through a portfolio approach. If returns aren’t important, then it should have been a grant. This is an absurd decision by the government,” he said.

“Why isn’t it going to be open-source?” was the most pressing concern. If it’s for Bharat, funded by taxpayers, shouldn’t it belong to Bharat? Industry veterans like Paras Chopra and Gaurav Aggarwal raised this question. Interestingly, Chopra has also submitted a proposal to the IndiaAI Mission, and Aggarwal is building AI infrastructure at JIO.

Although public interest use cases appear to be a priority, the risk of government influence on strategic decisions remains.

“Once you have the government as a shareholder, you never know how constrained your business decisions might become,” Mohandas Pai, chairman of Aarin Capital and former CFO at Infosys, told AIM. “Tomorrow, you raise capital—you’ll need government approval. You want to exit—you could face harassment or delays. That’s not how innovation ecosystems work.”

Pai argues that equity, if taken, should be managed through an investment fund—not held directly by a government department. “A fund acts like an investor. It understands how to manage a portfolio, negotiate fair terms, and support subsequent funding rounds. Government departments don’t,” he said.

“The key isn’t just about this GPU-for-equity round. The key is what happens in subsequent rounds. If the structure scares away future investment, then we’ve failed to build a sustainable ecosystem,” Pai added.

Access to GPUs, even with government subsidy, is a massive cost burden. Industry estimates suggest that 2,000 GPUs for one year would cost anywhere close to $8 million, depending on usage, even at Meity’s rates. Later, Moneycontrol reported that the Indian government is fully subsidising compute infrastructure costs for companies working on foundational AI models.

Importantly, this 100% subsidy is exclusive to core model training. Other GPU-heavy workloads—such as inferencing and application-level development—remain under the previously announced 40% subsidy scheme, according to the report. 

This clarification matters, as there had been confusion over whether foundational model training was also included in the 40% bracket.

To be clear, Sarvam will get access to around 4,000 NVIDIA H100s for six months from a pool of 14,000 GPUs. The rest will be distributed among Soket, Gnani, and Gan AI. Still, not everyone is buying the narrative of long-term enablement. “Government taking equity for giving access to GPU capacity using taxpayer money is not a good idea,” Pai said.

The Other Side of the Coin

According to Vivekanand Pani, co-founder and CTO of Reverie Language Technologies, the entire framing is wrong to begin with. “What we needed was not an AI model, but a national mission for digital language engagement—in schools, in governance, in business, in agriculture,” he said.

According to Pani, equity doesn’t ensure accountability. “A VC takes equity and puts pressure on the board. The government won’t do that. So what’s the accountability? Until value is created for Indian users, this is still public money used speculatively,” he said.

The real issue, he argues, is that India is racing to achieve LLM parity with the West, without first solving the problem of language access for its own people.

On the other hand, and in agreement with Upperwal from Soket AI, Nikhil Malhotra, chief innovation officer at Tech Mahindra, told AIM that for some startups, the equity trade-off makes sense. 

“It’s important that the government promotes technologies built by our own people, instead of always relying on external solutions. Standing behind companies like Sarvam shows belief in Indian innovation. As long as the equity doesn’t interfere with their direction but helps elevate them globally, it’s a fantastic idea,” Malhotra said.

One of the biggest issues facing Indian AI startups is the lack of enterprise support. Malhotra suggests that the government can step in—not just with funding but with legal and operational backing that gives AI startups room to persist.

Malhotra sees the government’s move as a positive step, so long as the equity does not limit startup autonomy. “Equity doesn’t necessarily mean just financial ownership. If the intention is to take Indian innovation to the world, then it’s a great decision by the government,” he said. 

The post IndiaAI’s Equity in AI Startups is a ‘Fantastic’ but Risky Idea appeared first on Analytics India Magazine.

Generated by RSStT. The copyright belongs to the original author.

Source