Claude's in-country inference in India officially launched on October 5, 2026: Anthropic announced that day that Claude Opus 5, Claude Sonnet 5, and Claude Haiku 4.5 are now available through Amazon Bedrock in India with in-country inference, meaning requests sent through Indian endpoints are processed directly on servers inside India, with no need to route them abroad. For heavily regulated customers such as banks, insurers, and government agencies, this step fills in not model capability, but the long-missing piece of data residency.
What has held back regulated industries was never that the models were not strong enough
Irina Ghose, Anthropic's Managing Director for India, put it plainly: in August this year the company promised to bring Claude inference to India, and it has now delivered; regulated industries and government customers want to use Claude for their most critical work, provided requests can be processed inside the country, and as of today they can. Sandeep Dutta, President of AWS India and South Asia, offered the other side of the same judgment: for these industries, data residency determines whether AI stays at the pilot stage or can move into production. Deploying Claude in India through Bedrock also comes with the audit trails and access controls that risk and compliance teams must check before going live. During the private preview, institutions such as Reliance and CRED had already begun testing it; Mohit Kapoor, Chief Technology Officer of the Mahindra Group, specifically pointed to India's Digital Personal Data Protection Act (DPDP Act) — once data sovereignty became a hard requirement, keeping workloads inside the country is what gives enterprises the confidence to roll Claude out across the entire group.
Why India is worth a dedicated set of endpoints from Anthropic
India is one of Claude's largest markets worldwide, and its developer community also ranks among the most technically dense. Anthropic set up an office in Bengaluru in February this year; when the Claude Partner Network launched in March, registrations from India ranked among the highest globally, and in August the company held its first partner certification there. The customer list underlines the same focus: banks such as Kotak, Axis, and IndusInd use Claude for engineering and employee productivity, the National Payments Corporation of India (NPCI) is building an agent platform called AiNxt on top of Claude, TCS plans to roll Claude out to 50,000 employees, and Infosys has set up a dedicated Anthropic Center of Excellence. Claude has recently appeared frequently on third-party platforms — for example, Google Antigravity launches Claude 5.5 follows a subscription distribution route; India's launch, by contrast, follows an infrastructure route — the model stays the same, what changes is where requests are processed.
The limits should be clear, too
In-country inference solves data residency; it does not mean compliance is automatically achieved: enterprises still have to handle data classification, retention periods, and internal authorization themselves, and the endpoint merely crosses "data leaving the country" off the checklist. Moreover, OpenAI and Google already offered local data processing in India before this, so Anthropic is closing a gap with its rivals rather than securing an exclusive position. The real point to watch comes next: once all three frontier model providers meet residency requirements in India, banks and government agencies will judge their choices by model capability, price, and the ecosystem itself, and the scale at which pilots convert to production will be the measure of how much this launch truly matters.
