Sarvam AI Raises $234M to Build India's First Sovereign AI Stack

India just minted its most consequential AI unicorn. Sarvam AI has raised $234 million in the first close of a $300 million Series B round at a post-money valuation of $1.5 billion, making it the largest pure-play AI funding deal in India's history. The round is anchored by a single, striking commitment: HCLTech will invest $150 million as the lead strategic investor, with Bessemer Venture Partners joining alongside existing backers Khosla Ventures and Peak XV Partners.

The company behind the raise is not a wrapper on GPT-4 or a fine-tuned Llama. Sarvam AI was founded in 2023 by Vivek Raghavan and Pratyush Kumar, the minds behind AI4Bharat and Aadhaar's biometric infrastructure at IIT Madras. Their thesis from day one: India needs AI built from scratch, on Indian soil, in Indian languages, for Indian scale.

A unicorn built on a different kind of moat

To understand why this round matters, you need to understand what Sarvam actually ships. Sarvam is the only Indian company building a complete vertical stack, from raw foundation models and speech systems to enterprise APIs and consumer applications. That full-stack ambition is now backed by real production numbers:

• Sarvam's conversational platform handles over 2 million interactions a day, with usage doubling in the last two months.
• Its inference platform processes 10 million API calls daily, with usage tripling in the last three months.
• Sarvam Vision is being used to digitize over 35 million pages, from insurance forms to legacy land records.
• Its speech models transcribe over half a million hours of audio each month.

The deployments are not just impressive in scale, they are genuinely novel in scope. With multilingual voice agents, Sarvam's platform collected high-quality data from 17 million farmers, providing deep insights to the Ministry of Agriculture and Farmer's Welfare. For one of India's leading insurance providers, a nationwide voice campaign supported low-cost policy renewals for 45 million policyholders.

The models under the hood

Sarvam's two flagship open-source models, Sarvam 105B and Sarvam 30B, are Mixture-of-Experts (MoE) architectures. MoE is a design where only a small subset of the model's total parameters are activated for any given input, keeping inference costs low while maintaining high capacity. Sarvam 105B has 106 billion total parameters with roughly 10 billion active per token and a 128K context window, while Sarvam 30B has 32 billion total parameters with about 2.4 billion active per token and a 65K context window.

Both are reasoning models trained from scratch on large-scale, high-quality datasets curated in-house across every stage of training: pre-training, supervised fine-tuning, and reinforcement learning. Training was conducted entirely in India on compute provided under the IndiaAI Mission. On agentic benchmarks specifically, Sarvam 105B's agentic index of 25 places it ahead of INTELLECT-3 and GLM-4.5-Air despite trailing both on overall intelligence.

A major focus during training was the Indian context and languages, resulting in state-of-the-art performance across 22 Indian languages for its model size. This is not a minor detail. The founders previously identified the "token tax" problem: it took 4 to 8 tokens to represent a single word in Hindi or Kannada, compared to just 1.4 for English, making AI prohibitively expensive and slow for Indian users. Sarvam's tokenizer was built to fix that.

Why HCLTech wrote a $150M check

The most strategically loaded part of this deal is the HCLTech anchor. India's legacy IT services giants have built empires on selling cheap engineering labor to Western enterprises. That model is under pressure. Roughly 30 to 40 percent of Indian IT services revenue comes from application development, maintenance, and testing, and analysts estimate productivity gains from AI could eliminate 9 to 12 percent of total IT services revenue over the next three to four years.

HCLTech's $150M is not just a financial bet, it is a hedge and a pivot. The investment buys influence over Sarvam's roadmap and a stake in India's sovereign AI story. By combining HCLTech's deep enterprise transformation expertise, trusted global client relationships, data and software IP, and engineering depth, the partnership would accelerate Sarvam's goal of building a powerful, end-to-end sovereign AI ecosystem for India and beyond.

The government backstory

This raise did not happen in a vacuum. Sarvam has been the Indian government's chosen champion for domestic AI infrastructure. The Government of India, under the IndiaAI Mission, selected Sarvam to build India's sovereign Large Language Model, and in a first-of-its-kind initiative, Sarvam received dedicated compute resources to build an indigenous foundational model from scratch.

Sarvam is one of 12 organisations selected under the Innovation Centre pillar of the IndiaAI Mission to develop indigenous foundational models, with financial and compute support amounting to ₹246.72 crore. In exchange for equity, the government provided subsidized access to 4,096 NVIDIA H100 GPUs at the Yotta data center. That compute subsidy was the fuel that let a three-year-old startup train a 105-billion-parameter model from scratch.

When the Union Cabinet approved the IndiaAI Mission in March 2024 with an investment of ₹10,372 crore over five years, it created the foundation that companies like Sarvam needed. The Series B is private capital following public infrastructure, a pattern that mirrors how UPI created the rails for India's fintech explosion.

Who wins, who watches nervously

The clearest winners are Indian enterprises and government bodies that want AI without data leaving the country. For citizens, this means interacting with AI that feels familiar, not foreign. For enterprises, this means unlocking intelligence without sending their data beyond borders.

The more interesting competitive dynamic is global. The global AI landscape has been dominated by OpenAI and Anthropic in the US, DeepSeek and Alibaba in China, and Mistral in France. India, despite its massive technical talent pool and rapidly growing AI market, has been largely absent from the frontier model conversation. Sarvam is the most credible attempt yet to change that. For OpenAI, Google, and Microsoft, India has been a distribution market. Sarvam is trying to make it a production market.

The pressure also lands on other Indian AI players. Other notable contenders include Krutrim and the government-backed Bhashini initiative. A $1.5B valuation and a government mandate create a significant gap to close. Indian AI startups raised approximately $3.94 billion in Q1 2026 alone, signaling a massive pivot toward high-stakes infrastructure and sovereign technology.

What comes next

The $234M announced today is the first close of a $300M round, meaning more capital is still expected to come in. The investment will fund Sarvam's continued research on training its next frontier model for agentic, coding, and cybersecurity use cases, as well as access to compute at scale to expand its forward-deployed motion across key verticals.

For builders and enterprises evaluating the stack, the practical question is what this unlocks. With the foundation of end-to-end capability across data, training, inference, and product deployment in place, Sarvam says it is ready to scale to significantly larger and more capable models, including models specialized for coding, agentic, and multimodal conversational tasks. Both the Sarvam 30B and 105B models are already available via API and open-sourced under Apache 2.0, so you can start experimenting today without waiting for the next model generation.

The deeper bet here is geopolitical as much as technical. Sovereign AI, the idea that nations and enterprises need AI infrastructure they control end-to-end, is rapidly moving from a policy talking point to a procurement requirement. As Sarvam co-founder Vivek Raghavan has argued, real AI sovereignty is not just about building models. It requires control over the entire stack, right from data and algorithms to compute infrastructure and hardware. With $234M, a government mandate, and India's largest IT company as a strategic partner, Sarvam now has the resources to attempt exactly that.