BENGALURU: Homegrown AI startup Sarvam has raised $234 million in a funding round led by HCLTech and Bessemer Venture Partners, catapulting its valuation to $1.5 billion and making it one of India’s newest AI unicorns.The round also saw participation from existing investors Khosla Ventures and Peak XV Partners. As part of the transaction, HCLTech invested $150 million as the lead strategic investor, acquiring a 10.4% stake in the company.The funding comes amid growing investor interest in India’s AI ecosystem. Earlier this year, AI cloud platform Neysa raised $1.2 billion in a Blackstone-led funding round to expand large-scale AI infrastructure.Sarvam co-founder Pratyush Kumar said the fresh capital will expand compute infrastructure, help hire AI talent and build new products, including an AI-assisted coding platform aimed at competing with global software engineering assistants such as Claude and Codex.
Focus on expanding compute infra
He said HCLTech’s investment would help Sarvam accelerate its international expansion while continuing to work with a broad ecosystem of enterprises and technology partners. HCLTech CEO C Vijayakumar said the partnership would combine Sarvam’s AI research capabilities with HCLTech’s global reach to create a full-stack AI platform for enterprises and govts. The Bengaluru-based startup’s revenue has grown rapidly, rising from rather negligible levels in FY24 to Rs 1.5 crore in FY25 and Rs 45.1 crore in FY26, according to unaudited financials, reflecting increasing adoption of its AI offerings.Kumar said Sarvam has demonstrated that India can build competitive frontier AI models with strong Indian-language capabilities. The company’s next phase will focus on end-to-end products in agentic AI and coding AI.“We want enterprises to be able to deploy trusted AI agents within their own control environments,” he said, adding that the company is also scaling its Voice AI and Document AI offerings, which are seeing growing enterprise adoption.Sarvam launched a 105-billion-parameter foundational large language model (LLM) and a suite of tools designed for commercial use.Kumar also revealed that Sarvam now has access to one of the first Blackwell-based AI compute clusters operating in India. Nvidia’s Blackwell architecture is designed for accelerated AI training and inference workloads.“While we don’t own the data centre infrastructure, we manage the software stack required to run these systems efficiently. Having compute located in India is strategically important because it lowers costs and reduces dependence on overseas infrastructure,” he said.Addressing questions related to investors from foreign shores backing a company championing sovereign AI, Kumar said global capital and Indian innovation are complementary.
