Sarvam, a Bengaluru-based artificial intelligence startup, is dedicated to developing a complete stack sovereign AI platform. Sarvam AI is now part of the unicorn club. To reach this milestone, the company went on to raise $234 million at the first close of its Series B round, bringing its valuation to $1.5 billion. Indian tech service major HCLTech led the investment round with an initial investment of $150 million and a stake of 10.4% in the firm as a lead strategic investor.

The round was also led by Global venture capital firm Bessemer Venture Partners, along with other institutional backers Khosla Ventures and Peak XV Partners. The large investment represents a major step in the Series B fundraising that Sarvam anticipates in a total of $300 million, and marks a fundamental shift in investor confidence toward the use of indigenous, foundational AI models.

Full-stack sovereign AI strategy and foundation

This funding round represents a crucial moment in a heated debate about country-specific digital sovereignty and the reliance on international technology. With recent regulations in the United States requiring major AI labs to restrict access to their most powerful models to some users, nations that depend entirely on foreign AI infrastructure face strategic risks. 

Sarvam has become the flagship of India’s standalone AI development in this climate. The startup, founded by well-known AI4Bharat foray entrepreneur-at-large at IIT Madras, Pratyush Kumar and Vivek Raghavan, works on building AI models that speak the local Indian languages and interpret local documentation, providing cheap intelligence specifically for local infrastructure.

Sarvam is taking a full-stack, sovereign AI approach, as opposed to many spin-off artificial intelligence firms, which build thin software applications on top of an external third-party platform. The company’s architecture includes not only an independent foundation model training system but also local inference infrastructure, enterprise software applications, and agentic platforms that cover the entire development cycle. 

This control-to-control deployment will enable the company to provide in-country, highly secure deployment environments, thus ensuring sensitive enterprise and government data are processed safely without being affected by routing through overseas cloud systems. The company claims its ability to train large neural networks from scratch in India. 

The startup, earlier this year, introduced its first foundational in-house models called Sarvam 30B and Sarvam 105B to support 22 Indian languages. The larger 105-billion-parameter system is built for complex enterprise workloads, whereas the 30-billion-parameter system is optimized to run efficiently on standard consumer hardware. It has now obtained additional access to one of the first advanced Blackwell-built AI compute clusters in India, enabling it to oversee the software stack and efficiently operate these complex compute platforms, while reducing reliance on international data centres.

Multimodal tools and product roadmap

Sarvam’s sovereign Artificial Intelligence is being deployed in the real world, not only by many people and in numerous important applications, but at scale and in huge volumes. For the customer-facing part, the startup’s conversational AI platform handles over two million engagements per day, which have seen a double-digit increase in the last few months. 

It supports roughly 10 million API calls per day on its inference platform. The company’s multimodal tools are deeply baked into commercial business, with its specialized speech models transcribing over 500,000 hours of conversations per month. Its document AI services have been able to turn legacy documents, land records, and insurance forms into over 35 million pages of digital documents.

These systems are taking effect and making a measurable difference in the public and corporate sectors. The Ministry of Agriculture and Farmers Welfare is now using Sarvam’s multilingual voice agents to gather essential insight directly from 17 million farmers across the country. 

In financial services, a large nationwide voice campaign allowed an insurance company to help 45 million policyholders update their policies at minimal cost, and the company also worked with SBI Life Insurance to implement generative AI applications for customer engagement and sales support. Over 350,000 field-based sales workers are on the platform, supported and empowered by the prominent fintech company using Sarvam’s agentic AI platform.

The newly acquired capital will be dedicated to substantially increasing Sarvam’s internal computing infrastructure, hiring specialized AI talent, and speeding up the development of more advanced models. The company will spend on the development of the best possible technological tools that support agentic systems, cybersecurity measures, and automated coding, which will facilitate the integration of AI into organizations in a secure manner.

This product development roadmap fits well with the fast revenue growth of the company throughout the year, as can be seen from its unaudited financial records from zero in FY15 to ₹1.5 crore in FY16 to ₹45.1 crore in FY17. The partnership with HCLTech will be aimed at synergizing Sarvam’s expertise in research with the IT industry’s expansive network of business connections and international engineering powers. 

The partners will build a highly competitive AI ecosystem for banking, insurance, government technology, and defence clients globally by combining these elements. This synergy will enable Sarvam to penetrate international markets faster and support them as a national champion in technological capability.

Conclusion

Sarvam becoming an AI unicorn worth a billion dollars is a landmark for India’s deep-tech sector and its technological aspirations. The company raised $234 million, backed by a diverse funding base of global venture capitalists as well as notable domestic corporate investors, which not only provided it with the financial resources needed to scale its entire infrastructure but also enabled it to secure strategic partnerships.

The company’s emphasis on creating affordable, voice-enabled, and multilingual AI models speaks directly to the specific operational needs of Indian businesses and governmental bodies. With the new capital injection, the startup is poised to improve next-generation agentic products and grow its local compute power capabilities, proving that India can develop and nurture its own strong AI models that can compete favorably with foreign tech platforms.