Skip to content

Anthropic launches in-country Claude inference in India through Amazon Bedrock

Anthropic announced on October 5 that Claude requests sent through Amazon Bedrock's India endpoint are now processed inside the country, unlocking regulated industries for production AI.

By VibecodedThis 2 min read
Aerial view of the Bangalore skyline from MG Road in 2024
Gpkp, via Wikimedia Commons (CC BY-SA 4.0)

Anthropic on Monday flipped the switch on in-country inference for Claude in India: requests sent through Amazon Bedrock's India endpoint are now processed on servers inside the country's borders. The rollout covers Claude Opus 5, Sonnet 5, and Haiku 4.5, and it is aimed squarely at banks, government bodies, and large enterprises that could not run their most sensitive workloads through overseas inference.

The announcement, which carries a Bengaluru dateline, delivers on a commitment Anthropic made in August. "Our customers, especially in regulated sectors and government, want to use Claude on their most critical work, with requests processed in the country. From today, they can," said Irina Ghose, Anthropic's managing director for India, in the announcement.

Why data residency is the unlock

For India's regulated industries, where requests are processed is the parameter that decides whether AI stays a pilot or reaches production, in the words of Sandeep Dutta, president of AWS India and South Asia. India's Digital Personal Data Protection Act has made data sovereignty a business imperative, a point Mahindra Group CTO Mohit Kapoor echoed in the launch materials: keeping data and AI workloads within Indian borders lets the group scale Claude across its companies with confidence.

Reliance and CRED were among the early customers that tested in-country inference in private preview. Deployments through Bedrock ship with the audit trails and access controls that risk and compliance teams want before signing off on production use, which is the whole point: the controls are built into the service rather than bolted on afterward.

Developers are the core constituency

The launch lands in a market Anthropic says is its second largest for Claude.ai, and software development dominates the workload mix: 45.2% of work-related tasks that Indian users bring to Claude are software development, the highest share of any market Anthropic serves, according to the launch coverage. That helps explain the names in the launch.

NPCI is building AiNxt, its agentic AI platform, on Claude. Freshworks says it runs Claude Code across its engineering organization to compress routine build and review work. Swiggy CTO Madhusudhan Rao described Claude as one of the few AI tools that has changed how his engineering teams approach problem-solving. TCS is rolling Claude out to 50,000 associates, and Infosys has set up a dedicated Anthropic Center of Excellence.

Part of a bigger India push

Anthropic opened its Bengaluru office in February and held its first Claude Certification for Partners there in August, certifying up to 5,000 people. India posted some of the highest registration numbers anywhere when the Claude Partner Network launched in March.

In-region inference is not a new pattern on AWS: the company announced similar in-region Claude inference for Seoul and Singapore on September 30. For developers, the practical upshot is that the India endpoint now behaves like the others, same models and APIs, with the inference boundary drawn at the border.