Sarvam AI Secures $234 Million to Build Sovereign Trillion-Parameter Model

Bengaluru-based Sarvam AI has successfully closed the first tranche of its $300 million Series B funding round, securing $234 million. The round was bolstered by a $150 million strategic investment from HCLTech, pushing the company's valuation to $1.5 billion. This capital infusion is earmarked for the development of a foundational AI model featuring over one trillion parameters, designed to be trained entirely within India.

Announced at the Epoch 2026 developer conference, the project aims to achieve token sovereignty by reducing reliance on foreign infrastructure. By training the model locally, Sarvam AI intends to address a significant portion of India's growing AI computing requirements. The model is currently expected to go live within the next six months, marking a major milestone for indigenous large language model development.

This initiative aligns with the broader goals of the $1.25 billion IndiaAI Mission, which provides critical subsidies for over 38,000 GPUs. The government's 2026 policy framework is specifically designed to foster such high-compute projects, ensuring that Indian startups have the necessary resources to compete globally while maintaining data and infrastructure control within national borders.