HCLTech backs Sarvam with $150M as India targets sovereign AI

HCLTech leads $234 million Series B for India's first AI unicorn

Sarvam, a Bengaluru-based startup building full-stack AI models, infrastructure, and applications, raised $234 million at a $1.5 billion valuation on Monday. HCLTech (per TechCrunch), the IT services unit of conglomerate HCL Group, committed $150 million as lead strategic investor. Bessemer Venture Partners joined alongside existing backers Khosla Ventures and Peak XV Partners.

The company says it intends to raise a total of $300 million in Series B. This marks Sarvam's first major capital infusion since its seed and Series A rounds closed $41 million more than two years ago.

Sarvam's core focus is building models and inference infrastructure optimized for Indian languages and use cases. The startup has deployed models across government, financial services, insurance, and defense sectors. Current operational metrics (company-reported) include 2 million conversational interactions daily, 10 million daily API calls on its inference platform, 500,000 hours of audio transcribed monthly via its speech models, and 35 million pages digitized monthly through document AI systems.

Recent deployments show scale in practice. Sarvam's multilingual voice agents collected data from 17 million farmers for India's Ministry of Agriculture. A voice campaign for a major insurer supported policy renewals for 45 million policyholders. A fintech company uses Sarvam's agentic AI platform to support a sales force of more than 350,000 people (company-reported).

The real driver: AI access and control, not just funding

This round sits at the intersection of two pressures. First, capital concentration: high compute costs and limited access to training infrastructure have made it hard for Indian startups to build frontier models, leaving the field dominated by U.S. and China-based competitors. Second, geopolitical control. Last week Anthropic disabled access to its latest models (Fable 5 and Mythos 5) after a U.S. government order barred foreign nationals from using them, citing national security. That order crystallized what had been abstract: access to cutting-edge AI is a chokepoint.

HCLTech's strategic involvement is the telling detail. The company brings enterprise relationships, an engineering workforce, and software assets that let Sarvam commercialize models faster and wider than capital alone would. A dedicated buyer with distribution and trust inside Indian government and corporate procurement can compress the path from model to deployment.

India is already OpenAI and Anthropic's second-largest market by user adoption, driven by developer density and cost-conscious enterprise adoption. But it has produced almost no competitive foundation models. Sarvam is among a tiny handful attempting to build full-stack AI for Indian languages. If the company can reach price and inference speed parity with U.S. models while offering language and regulatory fit for India, it becomes the default for a market of 1.4 billion people. If it cannot, it becomes a regional wrapper around foreign models.

For infrastructure teams: plan for fragmentation

The Anthropic access suspension is a preview of a world in which U.S. export controls tighten further. Teams relying on a single vendor's models or inference—especially those working with international data or deployed in government—should benchmark Sarvam's latency and cost against OpenAI and Claude now, before regulatory pressure forces the switch.

Sarvam's models ship in 30-billion- and 105-billion-parameter versions (open-source, released earlier this year). Test them in your dev environment. Know what inference cost per 1,000 tokens looks like. Establish a fallback before you need one.