Indian foundation model startup Sarvam AI is in the final stages of closing a funding round of roughly $300 million to $350 million at a valuation of about $1.5 billion, according to Bloomberg and multiple Indian business outlets. The deal, reportedly led by Bessemer Venture Partners with participation from Nvidia, Amazon and Prosperity7 Ventures, positions Sarvam as one of the most heavily capitalised AI startups in India and likely the country's largest private startup funding event of 2026 so far.
A homegrown answer to OpenAI and Google
Founded in 2023 by AI researchers Vivek Raghavan and Pratyush Kumar, Sarvam has focused on voice-first systems tuned for India's linguistic reality. Its Sarvam-30B and Sarvam-105B models, trained on large-scale Indian language datasets, are designed to handle 22 Indian languages and serve sectors where language diversity and cultural context matter — government services, financial services and other regulated industries.
Investor interest intensified after the company unveiled the 30-billion and 105-billion parameter models at the India AI Impact Summit in February, which were trained in India rather than on Western infrastructure. Reports describe the round as a defining moment in what local press is calling India's "foundational AI race," where domestic labs are trying to build credible alternatives to OpenAI, Google and Anthropic rather than simply fine-tuning foreign models.
Part of a bigger Indian AI buildout
Sarvam's raise lands against a backdrop of unusually large infrastructure commitments to India. Microsoft's previously announced $17.5 billion India investment is now beginning to materialise, with a flagship Hyderabad data center reportedly going live this summer. Taken together, new sovereign-scale compute and a well-funded local model builder reduce India's dependence on US hyperscalers for both training and inference.
The participation of Nvidia and Amazon is particularly notable. For Nvidia, backing Sarvam extends its pattern of taking strategic stakes in regional model labs that can anchor demand for its GPUs. For Amazon, the investment reinforces AWS's positioning as the default cloud for Indic-language AI workloads.
Implications
If the round closes on the reported terms, Sarvam will have the balance sheet to pursue significantly larger training runs and a denser product roadmap across voice, text and agentic applications. That matters beyond India: a competitive, sovereign model ecosystem in the world's most populous country reshapes assumptions about where frontier AI work can credibly happen.
It also raises harder questions. Valuations at or above $1.5 billion for pre-revenue-scale foundation model companies continue to test investor appetite, especially as frontier labs in the US trade at valuations two orders of magnitude higher. Sarvam's commercial traction with Indian government and enterprise customers will be the real test of whether this round marks the arrival of a durable new category leader — or a richly priced bet on industrial policy.
