Anthropic has launched in-country Claude AI inference in India through Amazon Bedrock, allowing prompts and responses sent through its India endpoint to be processed on servers located within the country. The move is aimed at enterprises, financial institutions, government bodies and other regulated users that require sensitive AI workloads to remain within India.
Amazon Bedrock, AWS’s platform for accessing and deploying generative AI models, will provide the infrastructure for local access to Claude. Anthropic said Claude Opus 5, Sonnet 5 and Haiku 4.5 are now available with in-country inference through Amazon Bedrock in India.
Under the arrangement, a request sent by an Indian customer through the India endpoint is processed on servers located in India, according to Irina Ghose, managing director at Anthropic India. She said regulated-sector customers had sought the capability for their most critical work because of data residency and data-handling requirements.
“The controls that keep processing within the country are built into the service, and the audit trails and access controls that risk and compliance teams require before they put AI into production,” Satinder Pal Singh, director of solution architects at AWS India and South Asia, said, according to the report.
The launch comes as Indian companies expand their use of generative AI across software development, banking, payments, healthcare, digital services and public services. India is Anthropic’s second-largest market for Claude.ai, with software development accounting for 45.2% of work-related tasks that users bring to Claude in the country, Ghose said. The company said this was the highest share for software development among its markets.
Several large technology and consumer companies are already using Claude in India. TCS is equipping 50,000 associates with the model, while Infosys is combining it with its Topaz platform. Cognizant uses Claude in systems it builds for clients, and LTTS is embedding the model in its engineering platforms. Swiggy has made some services accessible through the Model Context Protocol, while Instamart allows customers to browse and purchase more than 40,000 products through prompts.
For these deployments, local processing addresses one part of the infrastructure and compliance requirements involved in putting artificial intelligence into production. The India endpoint is intended to provide local handling of requests, while AWS has highlighted audit trails and access controls for risk and compliance teams.
Anthropic opened its Bengaluru office in February, its second office in Asia after Tokyo. The company plans to expand its India team and local partnerships. It is also working with Karya to improve Claude’s training data across Hindi, Bengali, Marathi, Telugu, Tamil, Punjabi, Gujarati, Kannada, Malayalam and Urdu.
India recorded among the highest registration numbers when Anthropic launched its Claude Partner Network in March. About 5,000 people were certified at the company’s first Claude partner certification event in Bengaluru in August. Anthropic said the introduction of local inference could allow enterprise deployments that had been waiting for the capability to proceed.