Activate Invests in Sarvam AI's Series B Round

Activate, an early-stage AI venture fund, has announced an investment in Bengaluru-based AI startup Sarvam as part of the company’s ongoing Series B funding round, adding another investor to a round that is targeting $300 million.

The investment comes after Sarvam raised $234 million in the first close of its Series B round at a $1.5 billion post-money valuation in June. HCLTech led the first close with a $150 million investment, alongside Bessemer Venture Partners, while existing investors Khosla Ventures and Peak XV Partners also participated. Sarvam said it intends to raise a total of $300 million in the round.

Announcing the investment on X, Activate said it views AI as one of the largest technology shifts and believes India has an opportunity to play a leading role in the sector. The firm said that achieving that goal would require companies building foundational AI technology, research, and products tailored to India’s needs.

“Realising that vision requires ambitious founders building foundational technology, world-class research, and products designed for India’s unique needs. Sarvam has consistently demonstrated that ambition,” Activate said.

The firm added that it is partnering with Sarvam Co-founders Pratyush Kumar and Vivek Raghavan as they continue building the company. 

Sarvam has not disclosed the size of Activate’s investment or whether the latest capital forms part of the remaining money yet to be raised in the Series B round.

Sarvam is set to raise $75 million from investors including NVIDIA, Glade Brook, Gaja Capital, IndiGo Ventures, and several individual investors, according to regulatory filings with the Registrar of Companies (RoC), as reported by ET.

The funding is expected to support Sarvam’s work across AI infrastructure, model research, and enterprise deployments. The company previously said the proceeds would be used to train its next frontier model for agentic AI, coding, and cybersecurity applications, while expanding access to compute infrastructure.

The company recently hosted Epoch, its frontier AI conference in Bengaluru, where it revealed that it is working on a trillion-parameter model expected to launch within the next six months.

Sarvam also hired Devendra Singh Chaplot, a former researcher at xAI, Thinking Machines Lab and Mistral AI, as an advisor to accelerate work on its frontier AI models.

Founded in 2023, Sarvam develops foundation models, speech AI, document intelligence, and enterprise AI products. Earlier this year, the company released its Sarvam 30B and Sarvam 105B models trained in India. 

It says its conversational AI platform handles more than two million interactions a day, its inference platform processes over 10 million API calls daily, and its speech models transcribe more than 5,00,000 hours of audio every month.