Google expands data residency efforts in India with local processing of Gemini 2.5 Flash

He said the initiative is in response to growing customer requests to host the company’s latest AI models in India and is part of its ongoing investments in the country.

Google announced on July 23 that it will allow developers to run its latest artificial intelligence (AI) model, Gemini 2.5 Flash, entirely within India-based data centers, expanding its data residency efforts in the country.

This move is expected to help Google attract customers in sectors such as financial services, healthcare, and the public sector, which handle sensitive data and have strict data residency needs, as AI becomes more widely adopted.

Rival OpenAI had also recently announced data residency in India, and other Asian countries such as Japan, Singapore, and South Korea for ChatGPT Enterprise, ChatGPT Edu, and the API platform in May.

"India is a massive opportunity for us, and one of the key areas of focus is the public sector," Bikram Singh Bedi, Vice President of Google Cloud Asia Pacific, told Moneycontrol.

He said the initiative is in response to growing customer requests to host the company’s latest AI models in India and is part of its ongoing investments in the country.

"There are certain sectors where they are not comfortable with data not being in India...we've repeatedly heard our customers say, ‘Hey, we’d like your AI models or your cloud infrastructure to be in India,’" Bedi added.

Gemini 2.5 Flash is Google's first fully hybrid-reasoning model that gives developers the ability to turn thinking or reasoning on or off. Designed for fast, low-latency applications, the model also allows them to set “thinking budgets” to control how much the model reasons, helping them optimise across quality, cost, and latency. It was first introduced as an early preview in April and was made available to all developers in June.

In October last year, Google had announced that Gemini 1.5 Flash will be hosted in India, enabling organisations to conduct processing entirely within India. The company currently operates two cloud regions in India - Mumbai and Delhi-NCR, both of which have been empanelled by the Ministry of Electronics and Information Technology (MeitY).

Earlier this year, Google Cloud partnered with the Uttar Pradesh government to launch an AI-powered agricultural network called Uttar Pradesh Open Network for Agriculture.

The open network, powered by Google's Gemini and Beckn Protocol, will provide millions of farmers with one-stop access to essential services, including advisory services, credit, mechanisation, and market connectivity to sell their produce. Beckn Protocol is an open standard that enables interoperability and decentralised discovery of products and services.

Google also announced a strategic collaboration with the Andhra Pradesh government to promote AI adoption across the state in December 2024. This includes development and implementation of AI solutions, focusing on key sectors like healthcare and sustainability, while also enhancing their digital infrastructure, developing AI skills, and empowering the local startup ecosystem.

Bedi said the company is also working with the Indian government on the Integrated Government Online Training (iGOT) Karmayogi platform, which is being used by civil servants to upskill themselves.

Another key use case is around latency. "There are certain workloads — especially in the AI space — where low latency is important. In such cases, a delay from accessing something in a different geography may not allow the use case to function properly," he said.

Google made this announcement at its developer event, Google I/O Connect, in Bengaluru along with a slew of other major updates.

Gemma 3n, the company's latest open model, has also been released for developers. Bedi said it is engineered to deliver highly capable, private, real-time AI directly on-device, with support for low-end hardware with as little as 2 GB RAM. The model also supports over 140 languages, including six Indian languages.

Bedi stated that three of the four startups selected by the IndiaAI mission to build the next generation of Make-in-India AI models are using the latest Gemma model. These include Sarvam AI, Soket AI Labs, and Gnani.ai.

Google is also rolling out Firebase Studio, a cloud-based AI app builder that lets developers build and ship production-quality full-stack AI apps with the help of Gemini. It helps people with designing an app all using natural language, images, drawing tools, and screenshots as prompts.

Bedi said that Google has now integrated its Gemini 2.5 Pro model into Firebase Studio as well as into AI Studio, the company's free web-based developer tool designed for rapid prototyping and experimentation with generative AI models.