Sarvam AI Plans $74 Million NVIDIA-Led Series B Extension
  • News
  • Asia

Sarvam AI Plans $74 Million NVIDIA-Led Series B Extension

The pending tranche would value the Indian sovereign AI startup at about $1.51 billion

8/4/2026
Ali Abounasr El Alaoui
Back to News

Indian sovereign AI startup Sarvam is set to raise approximately $74 million, or Rs 698 crore, through an extension of its ongoing Series B round. The financing has not been publicly announced as completed; reports citing Registrar of Companies filings state that Sarvam’s board approved the issuance of 20,244 Series B preference shares and seven equity shares at an issue price of Rs 344,570 each. NVIDIA is expected to lead the tranche, which follows Sarvam’s $234 million Series B first close in June 2026.


Strategic Investment and Key Participants

NVIDIA is expected to invest Rs 238 crore, approximately $25 million, while Glade Brook Capital is set to contribute Rs 190.3 crore, or about $20 million. Gaja Capital is expected to invest Rs 95 crore, and Sanjay Kalra and Jyotika Kapoor are set to contribute Rs 50 crore, with IndiGo Ventures and additional angel investors and family offices also participating. Reported participants include Vrijesh Agarwal, KJ Trust, and AL Trust, although the available reports do not provide an individual investment amount for every investor.

The proposed tranche forms part of Sarvam’s broader Series B financing rather than a separate completed funding round. In June 2026, the company announced that it had raised $234 million in the first close of a planned $300 million Series B at a post-money valuation of $1.5 billion. HCLTech invested $150 million as the lead strategic investor, alongside Bessemer Venture Partners and existing investors Khosla Ventures and Peak XV Partners.

Valuation and Shareholding Dynamics

Following the proposed allotment, Sarvam is expected to be valued at approximately Rs 14,342 crore, or $1.51 billion, on a post-money basis. This represents a modest increase from the $1.5 billion post-money valuation reported at the Series B first close, rather than an unchanged valuation. The June financing made Sarvam a unicorn, but the new tranche should be described as pending until the allotment is completed or formally announced by the company.

After the allotment, NVIDIA is projected to hold a 1.66% stake in Sarvam, while Glade Brook Capital is expected to own 1.33%. HCL Technologies’ holding would be diluted to 9.95%, compared with the approximately 10.46% stake associated with its June investment. Co-founders Vivek Raghavan and Pratyush Kumar are expected to retain a combined 38.44% stake.

Sarvam’s Mission and Technological Focus

Founded in 2023 and headquartered in Bengaluru, Sarvam develops sovereign AI models, training and inference infrastructure, and enterprise applications tailored to Indian languages and use cases. The company serves enterprises, developers, and government organizations, with focus areas including banking, insurance, government technology, and defense. Its technology spans language, speech, vision, and document-processing systems designed and trained in India.

Sarvam has released models including Sarvam 105B and Sarvam 30B, along with Sarvam Vision for handwriting and Indian-language records. The company says Sarvam 105B is designed for knowledge, reasoning, and agentic tasks, while Sarvam 30B is optimized for deployment on consumer hardware. Its broader strategy is to build a full-stack sovereign AI platform that allows enterprises and public-sector organizations to deploy AI systems suited to India’s languages, infrastructure, and operating requirements.


The approximately $74 million NVIDIA-led extension would add major global and domestic investors to Sarvam’s ongoing Series B financing. However, the transaction should not be presented as fully secured or completed solely on the basis of the regulatory filings and media reports currently available. If finalized on the reported terms, the tranche would value Sarvam at about $1.51 billion and provide additional capital for its model research, computing infrastructure, and enterprise and government deployments.