Inside India’s AI Impact Summit

You’ve likely seen the headlines on the world’s largest AI gathering in India last week. The NYT wrote that the summit promised to be “a global coming-out party for India’s AI aspirations.” TIME reported on delegates’ sentiments around government “posturing.” And X was abuzz with news about logistical challenges as the who’s who of Silicon Valley descended on India’s megacity capital.

I was here on the ground in Delhi, and here’s what stood out to me from a tech and policy standpoint.

1. Localize to optimize: AI will adapt to local cultural context

In a session on AI and human potential, a top AI exec opened by asking the audience, mostly Indian locals, their favorite use cases for ChatGPT. The answers surprised even him: everything from high-level coding to astrology, life advice, and health-related queries.

We’ve all heard it: ChatGPT makes us all sound the same, and AI-generated content is becoming easily identifiable. But if ChatGPT’s 100 million weekly active users in India are embedded in a cultural context that’s significantly different from the West, how will AI adapt to that context over time? My guess: models will localize to optimize, absorbing user preferences through feedback loops of how people are actually using the model. As this happens, different cultural groups may develop AI “signatures” – distinct defaults in tone and references.

And as India and other non-Western nations build their own frontier models, trained on different languages and histories, use cases might continue to surprise Silicon Valley. A visually striking example of AI meeting local tradition came from Reliance’s Jio Intelligence Pavilion, which featured AI-powered holograms of mythical figures from the ancient Indian epic “The Mahabharata.” AI, but make it steeped in 2,500-year-old history.

2. There’s an ongoing race to win the AI market in India

For anyone building in AI, India is a massive opportunity. Currently the world’s most populous country and the fourth-largest economy, it accounts for 5.8% of Claude usage, second only to the US. India is also the second-largest market for ChatGPT, with 18-24 year olds making up about half of the platform’s user base in the country. Multilingual, mobile-first, and highly price sensitive, the Indian market presents unique challenges and opportunities for global AI players.

The AI labs are already responding to this opportunity. OpenAI launched OpenAI for India, an initiative to partner with leading Indian companies to expand AI in the country. It kicks off with a partnership with Indian multinational Tata Group to deploy 100 megawatts of sovereign, AI-ready compute capacity, with the potential to scale to 1 gigawatt. OpenAI also announced plans to open new offices in Mumbai and Bengaluru this year. And as we mentioned last week, Anthropic launched major initiatives in the country as well.

Watch for trends in multilingual and voice-first models, distribution partnerships with local incumbents, and India-specific pricing and local compute for lower latency and enterprise compliance. Because adoption is concentrated in a handful of cities and there’s plenty of users left to be had in the population of 1.46 billion, no winners are locked in yet.

3. India invests in its own AI future

India wants to build “sovereign AI,” the ability to develop and operate more of its AI tech domestically rather than relying on foreign providers. In a major AI infrastructure push, Indian conglomerate Adani Group committed to building a $250-billion AI infrastructure ecosystem over the next decade. Adani committed $100 billion to renewable-energy-powered data centers by 2035, with an expectation to catalyze an additional $150 billion across manufacturing, servers, and sovereign cloud services. India’s government also established the Startup India Fund of Funds 2.0, worth ~$1.1 billion, to mobilize venture capital across deep tech and startups.

Meanwhile, homegrown startups shipped models and made big announcements. Softbank-backed vibe-coding startup Emergent reached $100 million in ARR 8 months post-launch and shipped a mobile app to build software on-the-go. AI startup Sarvam released two open-sourced models, launched a chat app, and teased smart glasses. BharatGen, a government-backed AI consortium, released a model that supports 22 Indian languages.

How will localization and sovereign AI converge? As models adapt to India’s context, the players racing to win India might also have to play on India’s terms: local compute, local partnerships, and competing with a push to keep AI capability within India.

In other news

1. Anthropic launches Claude Code Security to help teams tackle software vulnerabilities (Anthropic)
2. OpenAI allegedly sets compute target of ~$600B by 2030 (CNBC)
3. Perplexity drops ads from its platform, citing user trust concerns (Financial Times)
4. Nvidia reportedly set to finalize $30B investment into OpenAI, replacing previous $100B commitment (Financial Times)
5. OpenAI announces partnerships with major consulting firms to help enterprises deploy AI coworkers (OpenAI)

4 new AI tools to try

1. Claude in PowerPoint – install here (Anthropic)
2. Gemini 3.1 Pro – options for developers, enterprises, and consumers (Gemini)
3. Lyria 3, music generation model – try in Gemini and in YouTube Shorts (Gemini)
4. Grok 4.20 – try here (xAI)

Recommended reading

1. Child’s play: Tech’s new generation and the end of thinking (Sam Kriss, Harper’s Magazine)
2. Intelligent AI delegation (Nenad Tomašev, Matija Franklin, and Simon Osindero, Google DeepMind)
3. Does socialization emerge in AI agent society? A case study of Moltbook (Ming Li, Xirui Li, Tianyi Zhou, University of Maryland and Mohamed bin Zayed University of Artificial Intelligence)

See you next week!