Google Cloud boosts AI growth with IIT Madras alliance

• Google Cloud expands local AI hardware in India to meet data-sovereignty and low-latency requirements.
• Partnership with IIT Madras/AI4Bharat launches Indic Arena to benchmark models on Indian languages, cultural use-cases.
• New India-specific AI capabilities include Document AI, Gemini batch support, and location-aware grounding via Google Maps.

Google Cloud is making a major move in India by expanding its AI infrastructure and strengthening partnerships to better serve local needs. The company says it is significantly increasing its computing capacity within India, built on its AI Hyper computer architecture and powered by the latest Trillium TPUs, so that enterprises, startups and public-sector organisations can train and deploy advanced AI models right in the country.

This expansion serves two linked goals: providing high-performance, low-latency AI services and ensuring data-residency and sovereignty for Indian users. By enabling the training and serving of models like the Gemini series locally, Google Cloud wants to give Indian customers full control over data and model processing avoiding reliance on overseas infrastructure.

Beyond hardware, Google is also introducing new AI capabilities tuned for Indian contexts. These include batch processing support for Gemini 2.5 Flash for large-scale tasks, a preview of Document AI to automate document workflows, and a “grounding” feature using real-time data from Google Maps so that models can deliver location-aware responses in India.

Also Read: PM Modi's Bhutan Visit Strengthens Energy, Trade, Cultural Ties

A key part of the strategy is ecosystem-building. Google, together with IIT Madras via its research centre AI4Bharat, has launched the platform Indic Arena, which allows users to anonymously evaluate and benchmark AI models across Indian languages and cultural contexts. Google is providing cloud credits and compute support to power this community-driven initiative.

According to Mitesh Khapra of IIT Madras, building AI systems tailor-made for India’s linguistic and cultural diversity is central to AI4Bharat’s mission, and having a neutral standard benchmark is essential.

In summary, this push by Google reflects a shift from just offering global AI tools in India to enabling India-specific, sovereign AI infrastructure and ecosystem built by Indians, for Indian use-cases. The message is clear: local compute, local context, and local control.