Sarvam AI, the Bengaluru company that describes itself as a full-stack sovereign AI company, raised $234 million in the first close of a $300 million Series B round, reported on 15 June 2026. The round values the company at $1.5 billion after the investment. HCLTech was the lead strategic investor with $150 million, and Bessemer Venture Partners joined, while existing investors Khosla Ventures and Peak XV Partners continued their support. The company has said the money will fund research on its next frontier model, compute at scale and the expansion of its work with enterprises and governments, and it described the round as one of the largest for an Indian AI startup.
The argument that Sarvam is among the leading Indian AI companies rests on its models and its usage. It has released Sarvam 105B, which the company says matches or beats larger reasoning models on knowledge, reasoning and agentic benchmarks, and Sarvam 30B, which is optimised for devices at the edge. Its conversational platform handles over two million interactions a day, a figure that has doubled in two months, and its inference platform processes 10 million API calls a day, up three times in three months. Its voice agents collected data from 17 million farmers for the Ministry of Agriculture, its speech models transcribe over half a million hours of audio a month, and its vision model is digitising more than 35 million pages of insurance and land records. It also supported policy renewals for 45 million policyholders of a leading insurer. The models support more than 22 Indian languages.
The counter-argument is that the benchmark claims come from the company and that the funding is only the first close. The round is $234 million of a planned $300 million, so the balance of $66 million has still to be raised, and the valuation of $1.5 billion is a post-money figure agreed with strategic investors, one of whom, HCLTech, is also a partner in the Odisha data centre project. A company that builds models needs a very large amount of computing power, and the amount raised is small next to the budgets of the largest global AI companies. The report mentions access to compute at scale as a goal and not as something already secured. There is also the usual uncertainty about whether usage numbers translate into revenue, since the report gives no figures for sales or profit.
The significance for India is the idea of sovereignty. Sarvam has argued that enterprises and governments should be able to own and run their own AI systems, and the Indian languages it supports reach many people who are poorly served by models trained mostly on English. The investment by HCLTech links one of the country’s large IT services firms to a model developer, and the partnership with the Odisha government for a data centre extends that link into infrastructure. If the approach works, it could give India an alternative to relying on foreign models for government and critical services.
Sarvam has raised a large round and built models and users at a fast pace. The next questions are how much it earns, how much compute it can secure, and whether its models can keep up with the global leaders.
Leave a Reply