When Sarvam AI’s co-founder Pratyush Kumar posted on X after his company’s unicorn announcement last week, what stood out in his message was how bullish he was on building frontier-class AI systems from India.
It was a pointed remark, perhaps having greater context since it arrives just days after the United States government restricted foreign nationals from accessing Anthropic’s most advanced models. But it also captured an argument that Sarvam has been building its entire existence around.
The Bengaluru-based startup, founded in 2023 by Kumar and Vivek Raghavan, announced a $234 million raise as the first close of a planned $300 million Series B round, valuing the company at $1.5 billion. The round was led by HCLTech, which committed $150 million for a 10.46 percent stake, with Bessemer Venture Partners, Khosla Ventures, and Peak XV Partners also participating. The raise marks a sharp jump from Sarvam’s previous valuation of roughly $196 million just a year prior.
What separates Sarvam from most of the AI companies making headlines right now is what it is actually building. Most AI businesses take existing foundation models from OpenAI, Anthropic, or Google and build applications on top of them. Sarvam is doing something considerably more difficult. It is building its own models, running the infrastructure needed to power them, and creating products for enterprises and governments, all under one roof. The company describes itself as India’s full-stack sovereign AI startup, and that is significant.
The company’s government pipeline tells part of the story. In April 2025, it was selected by the Indian government to build the country’s first sovereign Large Language Model under the IndiaAI Mission. In February 2026, it signed sovereign AI infrastructure MoUs with Tamil Nadu and Odisha.
At the India AI Impact Summit the same month, it unveiled two models trained entirely from scratch on Indian-language datasets, a 30-billion-parameter model and a 105-billion-parameter model, responding directly to earlier criticism that its development had been too reliant on adapting foreign models.
The timing of the funding round was not incidental. Sarvam’s unicorn status arrives at a moment when the case for sovereign AI has shifted from a policy preference to something closer to a national imperative. Global AI access is no longer guaranteed. Geopolitical pressure is reshaping who can use what, and countries that have not built their own AI stack are discovering the limits of dependence.
The numbers behind India’s AI opportunity only sharpen the argument. The country’s AI spending is expected to grow from around $7 to 8 billion in FY23 to $20 to 22 billion by FY27, making it one of the fastest-growing AI markets in the world. Sarvam’s conversational platform now handles more than 2 million interactions daily, a figure that doubled in two months. Its inference platform processes 10 million API calls per day, tripling in three months.
HCLTech’s decision to anchor the round with $150 million is itself a statement. The company’s CEO, C Vijayakumar, described the investment as a step toward building India’s trusted and globally competitive AI ecosystem. Bessemer’s Pankaj Mitra went further, arguing that India’s AI ambitions require not just indigenous foundation models, but a complete sovereign stack covering infrastructure, data, and applications.
That stack is precisely what Sarvam is trying to build. Whether it gets there will depend on execution, on the pace of model development, on enterprise adoption, and on whether India’s government commitments translate into scale. But the unicorn status is not just a valuation milestone. It is a signal that the market, and the money, believes the country needs someone to try.
