Sarvam AI, India’s homegrown artificial intelligence startup, is finalizing a funding round of $300–350 million — roughly ₹2,900 crore — at a valuation of $1.5 billion, according to a Bloomberg report published on April 2, 2026. The Bengaluru-based company is set to become the largest pure-play Indian AI unicorn ever. The round could close within days.
Bessemer Venture Partners is expected to lead the round, with Nvidia, Amazon, and Prosperity7 Ventures also participating. This is not a typical startup cheque — when the world’s most powerful chip company and the biggest cloud platform both back the same Indian AI startup in the same round, it means something. Founded in August 2023 by Vivek Raghavan — the architect behind Aadhaar’s biometric systems — and Pratyush Kumar, a former researcher at IBM and Microsoft, Sarvam has gone from zero to unicorn in under three years.
What This Means for India
India has long been a consumer of AI built in California or Beijing. This round changes that story. Sarvam is not building a ChatGPT wrapper — unlike other Indian AI unicorns, it is training foundational models from scratch., on Indian soil, in 22 Indian languages. For a country of 1.4 billion people where most citizens speak neither English nor Mandarin, that is the entire point. Startups can now access Sarvam’s API to build voice-first apps in Hindi, Tamil, or Bengali. Enterprises get an AI vendor that understands Indian regulatory frameworks. And the government gets a sovereign AI partner it can actually trust with sensitive data — thanks to Sarvam’s newly launched Chanakya platform, built for air-gapped, on-premise deployment in defence and government environments.
Key Details at a Glance
- Funding: $300–350 million (approx. ₹2,900 crore), valuation $1.5–1.55 billion
- Investors: Bessemer Venture Partners (lead), Nvidia, Amazon, Prosperity7 Ventures
- Founders: Vivek Raghavan (Aadhaar architect) and Pratyush Kumar (AI4Bharat, IIT Madras)
- Products: Sarvam-30B and Sarvam-105B LLMs, Saaras V3 speech model, Chanakya (defence/govt platform), Indus consumer app — all supporting 22 Indian languages
- Govt backing: Selected under IndiaAI Mission; received ₹99 crore in GPU subsidies — 4,096 Nvidia H100 chips through Yotta Data Services, the single largest allocation under the mission
- Gujarat MoU: Signed March 18, 2026 to set up a Sovereign AI Park in Gandhinagar with advanced compute infrastructure, R&D centres, and youth skilling facilities
What Happens Next
The round is expected to close imminently, after which Sarvam will officially enter unicorn territory — a first for a pure-play Indian AI startup. The capital will go toward scaling the Chanakya platform across government and defence contracts, expanding the Sovereign AI Park in Gandhinagar, and continuing model development. At a time when India is racing to lead in AI, Sarvam’s next 12 months will likely define whether India produces a globally competitive AI company — or stays a footnote in someone else’s story.
FAQs
What is Sarvam AI and why is it called India’s sovereign AI startup?
Sarvam AI is a Bengaluru-based company building large language models and voice AI systems specifically designed for India’s 22 official languages. It is called “sovereign” because its models are trained entirely on Indian soil using Indian data, reducing dependence on US or Chinese AI platforms for critical government and enterprise applications.
Who are the founders of Sarvam AI?
Sarvam AI was co-founded by Vivek Raghavan, who spent nearly 12 years as the biometric architect of India’s Aadhaar system at UIDAI, and Pratyush Kumar, a former researcher at IBM Research and Microsoft Research who also co-founded the open-source AI4Bharat initiative at IIT Madras. Both previously led AI language research before starting Sarvam in 2023.
What will Sarvam AI do with the ₹2,900 crore funding?
The funds are expected to go into three main areas: scaling the Chanakya platform for defence and government deployments, building out the Sovereign AI Park in Gandhinagar, Gujarat, and expanding its foundational model research to compete with global players like Google Gemini and OpenAI on Indian-language benchmarks.