Sarvam AI

Overview

Bengaluru-based startup selected by the Indian government's IndiaAI Mission to build the country's sovereign large language model ecosystem, funded with GPU access from the Mission's compute pillar. Sarvam released Sarvam-105B in February 2026 — India's first fully domestically-trained, open-source 105-billion-parameter LLM — alongside a 32B mixture-of-experts model, both trained across 22 Indian languages. In June 2026 Sarvam raised $234 million in the first close of a $300 million Series B at a $1.5 billion valuation, with HCLTech and Bessemer Venture Partners joining existing investors Khosla Ventures and Peak XV Partners.

Dependency Map

One hop in each direction — see Strategic Connections below for the full, sourced breakdown of every relationship.

Strategic Connections

Strategic Risks

• Dependent on continued IndiaAI Mission GPU allocation for training compute
• Competing against well-funded global foundation-model labs with far larger compute budgets

Future Outlook

• Positioned as the anchor model-builder for India's sovereign AI strategy
• Series B proceeds expected to fund further scaling of domestically-trained models

Sources

• Sarvam AI (Jun 2026) — Sarvam raises $234M in first close of $300M Series B
• BuildFastWithAI (Feb 2026) — Sarvam-105B: India's Open-Source LLM for 22 Indian Languages