What Just Changed for AI in India?
OpenAI has announced that it will now offer local processing for certain API requests in India. This means that for eligible businesses, both the data sent to the model (the prompt) and the computation that generates the response (the inference) will happen
on servers physically located within India. Previously, while some data could be stored locally, the heavy lifting of AI processing often happened in data centres abroad. This new capability, termed 'inference residency', is a crucial distinction. It directly addresses a major hurdle that prevented many Indian companies, especially those in highly regulated sectors, from fully embracing generative AI for their core operations. According to Nitin Bawankule, OpenAI's Head of Enterprise in India, this move is designed to give businesses more control and remove a practical barrier to adoption.
The Power of Data Residency
For industries like banking, financial services, insurance (BFSI), and healthcare, data localisation is not just a preference; it's a regulatory mandate. Rules from bodies like the Reserve Bank of India (RBI) often require that sensitive customer information be stored and processed within the nation's borders. This has made many firms hesitant to use powerful cloud-based AI tools, as they couldn't guarantee that a customer's data wouldn't be sent to a server in Europe or the United States for processing. Local inference solves this problem directly. Now, a bank can use an OpenAI model to summarise internal reports or draft customer service responses with the assurance that the sensitive data involved never crosses India's digital borders. This shift from 'pilot project' to 'production-ready' is the single biggest impact of the announcement for enterprise India.
Understanding the 'Certain' Requests
The headline's use of 'certain' is important. This new capability is not yet available for all of OpenAI's models. At launch, the local processing is specifically for the GPT-6 Luna model, and it is accessed through the company's Responses and Chat Completions APIs. This covers a wide range of text-based tasks like analysis, summarisation, and content generation. To use it, developers must create a new API project and explicitly select India as the region for both data storage and processing, then use a specific regional endpoint. There are also other conditions. The service is available for customers on a standard pay-as-you-go plan who have active Zero Data Retention (ZDR) status, and it comes with a 10% price premium over standard processing.
Broader Strategy and Performance Gains
Beyond data security, running AI workloads locally has a significant performance benefit: lower latency. Latency is the delay between sending a request and receiving a response. When data has to travel to a server on another continent and back, that delay can be noticeable, making applications feel sluggish. By processing requests within India, this travel time is drastically reduced, leading to faster, more responsive AI applications. This announcement is part of a larger 'OpenAI for India' initiative, which includes partnerships with local firms like the Tata Group to build out domestic AI data centre capacity. It also follows similar moves by cloud providers like Microsoft Azure and Amazon Web Services, which have been expanding their in-country AI offerings to meet surging demand.
















