Google CEO Sundar Pichai's public endorsement of Sarvam AI at India's AI Impact Summit in February marks a significant moment for India's homegrown AI ambitions. Pichai told the audience he was "impressed" by the three-year-old Bengaluru startup's progress in building foundational AI models designed specifically for Indian languages and real-world use cases. For a company founded just three years ago, that validation from one of the world's most influential tech leaders carries real weight.

Sarvam AI's rapid ascent reflects a broader shift in how India approaches artificial intelligence development. Rather than relying entirely on global AI platforms built by companies like Google, OpenAI, and Meta, India is now investing in native AI infrastructure designed from the ground up for Indian users. The company just reached unicorn status, raising $234 million in the first close of a $300 million Series B funding round at a $1.5 billion valuation, led by HCLTech with a $150 million investment.

What Problem Is Sarvam Actually Solving?

The core issue Sarvam identified is deceptively simple but profound: global AI models struggle with Indian languages and scripts. When you feed a large language model text in Hindi, Tamil, Bengali, or other Indic languages, the system becomes inefficient, slow, and expensive for Indian users. Rather than patching over this limitation, Sarvam rebuilt the entire architecture of its AI models from the foundation to handle Indian languages natively.

At the India AI Impact Summit in February 2026, Sarvam launched a new generation of large language models, including models with 30 billion and 105 billion parameters using a mixture-of-experts architecture. The company also released text-to-speech, speech-to-text, and vision models. The 105-billion parameter model, named Indus, became available on the App Store, Google Play, and the web starting February 20.

What makes these models distinctive is their ability to understand regional accents, slang, and cultural nuances that global models routinely miss. Users can seamlessly switch between English and Indic languages like Hindi, Tamil, or Bengali without losing context or accuracy. Sarvam also developed Sarvam Vision, an optical character recognition model that scored 84.3% accuracy on olmOCR-Bench, surpassing Gemini 3 Pro and DeepSeek OCR v2, and achieved even higher accuracy of 93.28% on OmniDocBench v1.5.

How Is Sarvam Building Its Business Beyond Software?

Sarvam's ambitions extend beyond language models. The company announced partnerships with Qualcomm for generative AI solutions and with German company Bosch to integrate AI onto car panels. Most notably, Sarvam introduced Kaze smartglasses, its first hardware product, which supports more than ten Indian languages. Prime Minister Modi tested the device at the summit's expo, signaling government interest in the company's vision.

The company's founders bring deep expertise to this mission. Vivek Raghavan holds a PhD from Carnegie Mellon and previously served as vice president at a chip design firm. He spent twelve years at India's Unique Identification Authority helping build Aadhaar's biometric systems. Pratyush Kumar spent years building open-source AI datasets for Indian languages before co-founding Sarvam.

Why Does Sarvam's Timing Matter for India?

Sarvam's unicorn funding arrives at a critical moment for India's technology independence. When the U.S. government ordered Anthropic to restrict foreign access to its most advanced AI models, companies outside America received a stark reminder that their AI infrastructure depends on foreign goodwill. For Indian banks, insurers, and government departments that had been building applications on foreign AI APIs, this restriction highlighted policy risk.

HCLTech's $150 million investment in Sarvam provides something many AI model companies struggle to secure: direct enterprise distribution routes into large customers in banking, manufacturing, and telecom that already trust the seller. This combination of native AI capability and trusted enterprise access creates a compelling value proposition for India's digital economy.

Steps to Understanding Sarvam's Competitive Advantage

• Language-Native Architecture: Rather than adapting global models to Indian languages, Sarvam built its entire system from the foundation to process Indic scripts efficiently, reducing computational costs and latency for Indian users.
• Cultural and Linguistic Nuance: Sarvam's models understand regional accents, slang, and cultural context that global models miss, enabling seamless code-switching between English and Indian languages without losing meaning.
• Enterprise Distribution Network: HCLTech's investment provides direct access to large enterprises in banking, manufacturing, and telecom, reducing the go-to-market friction that typically challenges AI startups.
• Government Support and Infrastructure: The Ministry of Electronics and Information Technology selected Sarvam as one of companies to develop an indigenous foundational model under the IndiaAI Mission, providing access to computing infrastructure and GPUs for model training.

Sarvam's rise also reflects India's broader AI policy shift. In December 2023, the company announced seed and Series A funding of approximately $41 million, led by Lightspeed Venture Partners, with participation from Peak XV Partners and Khosla Ventures. Then in April 2025, the Ministry of Electronics and Information Technology selected Sarvam as one of the companies to develop an indigenous foundational model under the IndiaAI Mission, providing access to computing infrastructure including GPUs allocated for model training.

Pichai's endorsement at the India AI Impact Summit signals that even as Google competes globally with other AI giants, the company recognizes the strategic importance of supporting India's homegrown AI ecosystem. For Sarvam, the validation from Google's CEO provides credibility with enterprise customers and investors who might otherwise view the startup as a regional player rather than a serious contender in the global AI race.