Google Launches Sovereign Gemini for Banks and Government in India

Google announced at the I/O Connect India 2026 conference in Bengaluru that Indian companies in the banking, healthcare, telecommunications and public administration sectors can now run the Gemini Flash and Gemini Enterprise models entirely on infrastructure physically located in India. The move is meant to address growing data sovereignty requirements in regulated industries.

Until now, Indian companies using Gemini models had to accept that their queries and data would be routed to Google data centers outside the country's borders. The new offering keeps both the data and the model inference itself within India, which matters as data localization rules tighten in sensitive sectors.

What Google Is Changing

Gemini 3.5 Flash is being rolled out to Indian enterprises and startups through the Gemini Enterprise Agent Platform and the Gemini Enterprise app, with model processing happening entirely within the country and no need to send data abroad. Google describes this as a strict commitment to keeping machine learning processing within national borders.

For banks, this means they can deploy AI agents for customer service or risk analysis without violating internal policies requiring customer data to stay within Indian territory. Hospitals and telecom companies can expect similar guarantees when processing medical or network data.

Investment in Andhra Pradesh

The sovereign Gemini announcement coincided with Google confirming a roughly $15 billion investment, spread across 2026-2030, in the company's first AI hub in India, located in Visakhapatnam, Andhra Pradesh. The project includes a gigawatt-scale data center, new energy sources and an expanded fiber-optic network.

This AI hub is a multidimensional investment combining a powerful gigawatt-scale data center, large-scale new energy sources and an expanded fiber-optic network. - Thomas Kurian, CEO of Google Cloud

The infrastructure will be built in partnership with local players AdaniConnex and Airtel. The project also includes a new international subsea cable landing station in Visakhapatnam, complementing existing cable landings in Mumbai and Chennai, intended to strengthen India's digital resilience for international connectivity.

Regulatory Context

India has spent years developing data localization rules, particularly in the financial and healthcare sectors, where regulators require sensitive citizen data to be stored within the country. American tech companies, including Google, Microsoft and Amazon, have long competed for the trust of Indian public institutions and regulated industries by offering increasingly localized versions of their cloud and AI services.

Google emphasizes that the new offering is also meant to support the government's Viksit Bharat 2047 program, India's long-term strategy to become a developed economy by the centenary of its independence. The company expects local AI hosting to speed up adoption in public administration, where concerns about data leaving the country have so far slowed procurement decisions.

What It Means Beyond India

Google's move fits into a broader trend in which large language model providers are offering more sovereign variants tailored to local regulations, rather than a single global cloud model. Similar questions about data localization and model inference are also emerging in the European Union, where the AI Act and data protection rules are pushing providers toward similar infrastructure concessions.

For Polish companies considering deployments of large language models in regulated sectors such as banking or healthcare, India's example shows that cloud providers are increasingly willing to host inference locally when market scale and regulatory pressure demand it. So far, Google has not announced a similar offering for Central Europe.