After Krutrim, which has somewhat fizzled out in the last few months, India has a new AI unicorn.
Bengaluru-based Sarvam announced on Monday that it has raised $234 million in the first close of its Series B round, taking its post-money valuation to $1.5 billion. The round isn’t fully done yet either — the company is targeting a total of $300 million for this Series B, which means another $66 million is expected to come in before the books close.
The round was led by HCLTech, which put in $150 million as the lead strategic investor. Bessemer Venture Partners joined as a new backer, while Khosla Ventures and Peak XV Partners, who had backed Sarvam since its $41 million seed and Series A round in 2023, returned to participate again. By most counts, this is the largest Series B round in Indian startup history.
Sarvam was founded in 2023 by Vivek Raghavan and Pratyush Kumar, who had previously worked together at AI4Bharat out of IIT Madras. The company has spent the years since building what it calls a full-stack sovereign AI offering for India — models, inference infrastructure, and deployment tooling meant to run without dependence on foreign AI providers. The fresh capital is going toward training its next frontier model, with the company specifically calling out agentic, coding, and cybersecurity use cases, along with buying more compute to scale up its enterprise and government deployments.
The traction numbers Sarvam has put out alongside the raise are fairly substantial. Its conversational platform is now handling over 2 million interactions a day, which it says has doubled in just two months. Its inference platform processes 10 million API calls daily, having tripled in three months. The company’s voice agents have also been used to collect data from 17 million farmers for the Ministry of Agriculture, and have supported policy renewals for 45 million policyholders at an unnamed insurer.
On the model side, Sarvam has put out two flagship releases recently — Sarvam 105B, a large reasoning model, and Sarvam 30B, built for edge deployment on consumer hardware — both trained from scratch rather than fine-tuned on top of an existing open-source base. Sarvam 105B has placed India sixth on the Artificial Analysis Intelligence Index, currently the only Indian entry on that leaderboard. The company has also built Sarvam Vision, a model for handwriting and Indian-language document recognition that’s reportedly being used to digitise over 35 million pages.
Much of this has been made possible by the government’s backing under the IndiaAI Mission, which gave Sarvam access to 4,096 Nvidia H100 GPUs in exchange for an equity stake — a structure that signalled just how seriously the government was willing to treat sovereign AI as a strategic priority rather than just an R&D grant. Sarvam has since gone on to use its models in places well outside the usual chatbot use case, including a recent tie-up with Pixxel to put its inference stack aboard an Indian satellite, and a deal with Meta to bring Hindi voice support to Ray-Ban Meta glasses.
The contrast with Krutrim is hard to miss. Ola founder Bhavish Aggarwal’s company had once raised at a billion-dollar valuation off the back of a flashy public launch, only for users to discover within hours that the model was telling people it had been built by OpenAI. Aggarwal himself later admitted that Krutrim was nowhere close to global benchmarks, with its models trailing even cheaply trained open-source releases from European labs by a wide margin. Sarvam, by comparison, took its time, kept its head down on research, and is only now stepping into the spotlight with a raise nearly one and a half times Krutrim’s peak valuation.
Sarvam co-founder Pratyush Kumar framed the round as a statement about India’s AI ambitions more broadly. “Country of India scale cannot rent intelligence. We have to build it ourselves. We are assembling the team, the compute, and the deployment engine to make this happen,” he said in a post on X. HCLTech CEO C Vijayakumar, for his part, described the investment as a step toward building a globally competitive AI ecosystem out of India, combining Sarvam’s research with HCLTech’s enterprise reach and engineering talent.
Whether Sarvam can hold on to this momentum the way Krutrim couldn’t will depend on what its next frontier model actually looks like once it ships. For now, India has its sovereign AI bet, backed by a domestic IT giant, a fresh war chest, and a valuation that puts it well ahead of anything an Indian AI startup has managed before.