Beyond Apps: The Foundational Shift
The conversation around AI in India has evolved. While many businesses integrate AI tools like chatbots or analytics, a new, more ambitious class of startup is emerging. These companies aren't just using AI; they are building the core infrastructure.
This includes creating large language models (LLMs) from the ground up, developing AI-specific data centres, and building platforms that allow other companies to deploy AI securely. This move marks a strategic pivot from being consumers of global AI technology to becoming architects of India's own AI future.
Meet India’s New AI Architects
This new breed of startup is defined by deep technological expertise and immense capital requirements. At the forefront are companies like Sarvam AI and Krutrim. Sarvam AI, founded by AI researchers Vivek Raghavan and Pratyush Kumar, recently raised $234 million in a funding round that valued it at $1.5 billion. The company is building a "full-stack" AI platform, from custom models trained on Indian languages and contexts to the infrastructure needed to run them. Similarly, Krutrim, founded by Ola's Bhavish Aggarwal, became the country's first AI unicorn in 2024. It focuses on creating LLMs that understand over 20 Indian languages, a crucial gap that global models often fail to address. These startups represent a bet on creating what is known as 'sovereign AI'—the capability for a nation to develop and control its own AI systems.
The Investor's New Playbook
Venture capitalists are taking notice and writing bigger cheques. The investment thesis is clear: owning the foundational layer of AI is a far larger opportunity than owning a single application. By funding the 'picks and shovels' of the AI gold rush, investors gain exposure to the entire ecosystem's growth. The recent funding rounds for Sarvam AI, which included major players like HCLTech, Bessemer Venture Partners, and Peak XV Partners, highlight this trend. HCLTech's strategic investment of $150 million into Sarvam underscores a belief that combining foundational AI technology with enterprise-level distribution is a winning formula. This isn't just about financial returns; it's a strategic move to ensure India has control over critical technology infrastructure, reducing dependency on foreign providers.
Government as a Catalyst
This startup boom isn't happening in a vacuum. The Indian government is playing a pivotal role through the IndiaAI Mission, a comprehensive national strategy backed by an outlay of over ₹10,300 crore. The mission aims to bolster the ecosystem by democratising access to computing power, improving data infrastructure, and providing risk capital for startups. A key component is making a massive pool of 10,000 or more Graphics Processing Units (GPUs)—the specialised chips essential for training AI models—available to Indian researchers and startups. Furthermore, the government is exploring a dedicated fund of up to ₹20,000 crore to provide long-term risk capital for these frontier AI companies, acknowledging that the cost of building advanced models can be prohibitive for traditional venture funding.
