LLMs are a distraction, India needs AGI for geopolitical edge, says Fractal's Srikanth Velamakanni

Srikanth Velamakanni , Co-Founder, Group Chief Executive & Vice-Chariman_ Fractal

India should prioritise developing artificial general intelligence (AGI) over addressing language diversity through large language models (LLMs), according to Srikanth Velamakanni, co-founder and CEO of AI startup Fractal and advisor to the IndiaAI Mission.

He said that just aiming to solve for the language diversity issue using Large Language Models (LLMs) is a “distraction” for the country.

Velamakanni argues that AGI—AI systems capable of matching or surpassing human intelligence across all domains—holds the potential for significant geopolitical advantages, particularly with the US and China.

Speaking to Moneycontrol, Velamakanni said, “All these existing large language models are inherently capable of handling most Indian languages…It is a distraction to think that it is about language and Indian values. It’s actually about AGI.  We need AI that can accomplish tasks in the real world. If these AIs can get really good, it creates a geopolitical advantage for companies and countries and therefore most countries are thinking to have their own strategy as well.”

“What if tomorrow US says, I can't export these models or it becomes a bargaining chip for any negotiation, we want to be careful that India is in a stronger bargaining position in these situations. That's why I think progress in AI is important and therefore whatever we're investing behind India's LLMs, these foundational models have to be thought of in the light of AGI progress,” he added.

The government's IndiaAI Mission is currently accepting proposals to support the development of indigenous foundational AI models. The mission has allocated a total grant of Rs 1,500 crore for this and has received around 67 applications so far. Fractal plans to submit a proposal to build sovereign LLMs under this initiative.

Despite being a significant step, Velamakanni believes the investment is insufficient. He noted that the Rs 1,500 crore grant for LLM development and the overall Rs 10,000 crore budget for the IndiaAI mission was "too little."

“I still think that India is spending too little. We announced Rs 1,500 crore and it looks like a large number.  But Rs 1,500 crore is about $180 million, that's the amount of money that the Canadian government has invested in one company. Canadian government has made some $2 billion dollars of grants overall… We need to up the ante to something close to a billion dollars every year, that's what I would do, but that's from my lens because I'm a tech guy,” he said.

Earlier this month, Fractal completed 25 years since it was launched, the AI unicorn will be filing its DRHP (Draft Red Herring Prospectus) soon as it plans to get listed this year.

Confirming the same, Velamakanni shared, “So the size of the IPO will definitely be over $500 million. We are not looking at substantial offloading at the time of the IPO. See the minimum required is 10% shares, so we will be as close to that as possible.”

An industry sourced had told Moneycontrol, that the company might file its DRHP in the next couple of months. Without disclosing any details or a timeline, Velamakanni confirmed, “Hopefully.”

While he did not confirm on the valuation the company is aiming for, according to previous media reports, Fractal may be seeking a valuation of $3.5 billion post IPO. The IPO may see both new and existing shares in the offering.

Fractal, dual-headquartered in Mumbai and New York, was co-founded by five graduates of the Indian Institute of Management Ahmedabad in 2000. The startup turned a unicorn, valued at over $1 billion, in 2022.

Three of the founders have since exited. Co-founder and group Chief Executive Officer Velamakanni and co-founder and Fractal CEO Pranay Agrawal own about 10 percent shares each.

The company backers include private equity firms TPG Inc, Khazanah Nasional, TA Associates and Apax Partners.

The company has over 5,000 employees in 18 locations including the UK, Singapore, Middle East and Australia, as well as India and the US. Its clients include Alphabet Inc.’s Google, Unilever Plc and Kraft Heinz Co.