India has quietly become the most important frontier market for artificial intelligence. In early 2026, a parade of tech CEOs descended on New Delhi for the India AI Impact Summit, collectively pledging tens of billions of dollars into the country's AI ecosystem. At the same time, a homegrown startup called Sarvam AI unveiled frontier-class models built specifically for Indian languages and culture. Together, these developments signal something bigger: India is no longer just a consumer of AI, it is becoming a builder.

The investment flood

The numbers are staggering. At the India AI Impact Summit in February 2026, global tech companies committed over $70 billion in AI-related investments in India. Microsoft announced it was on pace to invest $50 billion in AI across the Global South by the end of the decade, with India as a centerpiece. This followed an earlier $17.5 billion commitment in late 2025 to scale cloud and AI infrastructure across cities like Bengaluru, Hyderabad, and Pune. OpenAI and AMD both struck partnerships with Tata Group to build AI capabilities on Indian soil. Blackstone participated in a $600 million equity raise for Neysa, an Indian AI infrastructure company. And perhaps most eye-catching, Adani Group pledged $100 billion to build renewable energy-powered AI data centers by 2035, projecting a $250 billion AI infrastructure ecosystem over the decade. A delegation of 120 American chief executives attended the summit, a clear signal that India's AI ambitions have the attention of global capital.

Why India, why now

Several forces are converging to make India uniquely attractive for AI investment. Scale of the digital population. India has over a billion internet users and roughly 700 million smartphone owners, making it one of the largest potential markets for AI services globally. OpenAI's Sam Altman noted that ChatGPT already has more than 100 million weekly active users in India. Talent density. India produces more engineering graduates than any country except China. Indian technologists already lead AI teams at every major Western tech company. The new investments create opportunities for that talent to build world-class systems domestically rather than emigrating. Adoption speed. According to a Deloitte survey, India has the highest generative AI adoption rate in the Asia Pacific, with 90% of students and 80% of employees actively using the technology. A Boston Consulting Group report put India's overall AI adoption rate at 30%, above the global average of 26%. Government alignment. India is offering a 20-year tax break on overseas revenue from global data services, along with a $1.1 billion state-backed venture fund. The India AI Impact Summit itself was effectively New Delhi's coming-out party as an AI power, with coordinated policy and private sector action rather than isolated corporate bets. Data access. For companies training large models, India's massive and digitally active population provides a rich, diverse dataset spanning dozens of languages, cultures, and use cases.

Sarvam AI and the case for sovereign models

Amid the flood of foreign capital, one of the most interesting developments is homegrown. Sarvam AI, an Indian startup based in Bengaluru, unveiled its new lineup of large language models at the summit in February 2026. The release includes two flagship models: a 30-billion parameter model and a 105-billion parameter model, both built from scratch for Indian languages. These are a dramatic leap from Sarvam's earlier 2-billion-parameter model released in October 2024. Both models use a mixture-of-experts architecture, which activates only a fraction of their parameters at inference time, reducing computing costs. The 30B model supports a 32,000-token context window aimed at real-time conversational use, while the 105B model offers a 128,000-token window for complex, multi-step reasoning. Crucially, these models support all 22 official Indian languages and are optimized for voice interaction, a critical feature in a country where many users prefer speaking over typing. Sarvam also released updated speech-to-text (Saaras V3) and translation (Mayura) models alongside the LLMs. The company positions itself as a sovereign AI platform, the idea being that India should not depend entirely on foreign AI systems for its digital infrastructure. Sarvam followed up the model launch with Indus, a consumer-facing AI chat app designed to compete with ChatGPT and Claude in the Indian market. This is not just nationalism for its own sake. Indian languages carry nuances, code-mixing patterns, and colloquial expressions that global models often handle poorly. A model trained specifically for this context has a real performance advantage for Indian users.

Infrastructure versus intelligence

Not everyone is convinced the current wave is transformative. Critics point out that data centers alone do not equal AI leadership. Apar Gupta, founding director of the Internet Freedom Foundation, has noted that data centers "add servers and storage, not the ability to build and control advanced AI." The distinction matters. Much of the announced investment is in physical infrastructure, servers, power, cooling. Building frontier AI models, developing novel research, and creating a self-sustaining AI innovation ecosystem require a different set of capabilities. India's strength lies in its talent pool and adoption rates, but converting those into homegrown research breakthroughs remains an open challenge. Sarvam's models are a promising start, but the gap between a strong multilingual LLM and the frontier capabilities of labs like OpenAI or Anthropic is still significant.

What makes this moment different

Previous waves of tech investment in India, the IT outsourcing boom of the 2000s, the startup surge of the 2010s, were driven primarily by cost arbitrage. Companies came to India because labor was cheaper. The AI wave is different in character. Companies are coming because India offers something they cannot easily get elsewhere: a massive, digitally active population that can serve as both a training ground and a market. The government is actively coordinating with industry rather than simply offering tax breaks. And Indian companies like Sarvam are building competitive technology rather than just providing services. Whether India becomes a true AI superpower or primarily an infrastructure and talent supplier remains to be seen. But the ingredients, capital, talent, adoption, policy support, and now homegrown model development, are all present in a way they have not been before.

Practical takeaways

References