Sarvam AI targets $300–350M round at $1.5B valuation with Nvidia and Amazon

Sarvam’s expected round would make it one of the first India‑native foundation model labs to reach unicorn scale purely on the back of sovereign AI positioning. The investor mix—Bessemer leading, with Nvidia and Amazon alongside Prosperity7—illustrates how global capital and compute providers are treating India not just as a market but as an AI power center in its own right. If the deal closes near the reported $1.5B valuation, Sarvam will have enough runway to keep training and serving large MoE models tailored to India’s languages and regulatory context, without immediately needing a Big Tech acquisition.

In the race to AGI, this matters because it diversifies where serious model R&D happens. India has already signaled sovereign ambitions via the IndiaAI Mission; Sarvam is becoming the default private‑sector vehicle for those ambitions in the LLM space. With agentic models and 20‑plus language support baked in, Sarvam is betting that voice‑first, vernacular interfaces are the right path for a 1.4‑billion‑person democracy, rather than simply localizing U.S. or Chinese stacks.

It also shifts the balance for hyperscalers. Nvidia gains a marquee Indian model partner that helps justify GPU exports and local data center builds. Amazon hedges against purely Western labs by backing a player that could anchor AWS‑based sovereign deployments. For OpenAI, Anthropic and Google, the message is that India’s next hundred million AI users may come through domestic brands first, with American labs increasingly relegated to underlying APIs or co‑branded agents.