Claude Now Runs Inside India for Regulated Enterprises

On Monday, 5 October, Anthropic said its Claude models are now available with in-country inference in India through Amazon Bedrock. Requests sent to the India endpoint are processed inside the country. The models named are Claude Opus 5, Sonnet 5 and Haiku 4.5.

Why this matters more than it sounds

For most Indian enterprises the model was never the hard part. Where the data goes was. Banks, insurers, government bodies and large enterprises working under strict data-handling rules have asked for local inference for a long time, and until a model runs in-country, many of them stay stuck in pilot mode.

Sandeep Dutta, president of AWS India and South Asia, put it plainly: data residency is the parameter that decides whether AI stays confined to pilots or reaches full production. That is the sentence I would pin on the wall of any AI steering committee.

What Anthropic and AWS are claiming

• Requests through the India endpoint are processed on servers based in India.
• Deployments through Bedrock come with the audit trails and access controls that risk and compliance teams ask for before production.
• Anthropic had committed to this in August. Irina Ghose, managing director for India, said the company is delivering on that promise and will keep growing its India team and local partnerships.

Anthropic also described India as one of Claude's largest markets globally.

My read

Residency answers one question on a compliance checklist. It does not answer the rest: model risk management, vendor concentration, audit evidence, what happens to logs. But it removes the objection that ended many conversations before they started. I expect regulated BFSI teams to reopen use cases they shelved last year, and the ones that move first will be the ones whose governance work is already done.

Source: Business Standard - Anthropic brings in-country Claude inference to India via Amazon Bedrock