India is one of the largest markets for Claude, home to a developer community doing some of the most technically complex AI work we see anywhere. A top ask from organizations in India has been for options to access Claude locally.
In the coming weeks, Claude will be available with in-country inference in India through Amazon Bedrock, Amazon Web Services’s platform for building generative AI applications and agents at production scale. Requests sent to Claude through the India endpoint will be processed on servers in India.
Alongside the addition of in-country inference, we are also sharing more details about our work in India—including new partnerships with organizations deploying Claude at scale.
Running Claude inferencing in-country gives Indian organizations the certainty they need to serve customers in highly regulated verticals. For banks, insurers, public and government agencies—institutions that set exacting standards for how customer information is handled, whether those standards come from their regulators or their own policies—that control is the difference between evaluating AI and deploying it across their full organization.
"We are glad to see partners like Anthropic and AWS bringing in-country inference to India. Access to trusted, frontier AI closer to home is a positive step for the ecosystem as we all explore how these technologies can responsibly serve customers", said Ashok Vaswani, Managing Director & CEO of Kotak Bank
In-country inference also comes with the supervision to match. Claude deployments through Amazon Bedrock carry the audit trails and access controls that risk and compliance teams require before anything reaches production.
“Bringing Claude inference to India is a turning point for enterprise AI in this country,” said Irina Ghose, Managing Director of India, Anthropic. “India’s banks, insurers, telecoms, public and government agencies steward the data of a billion people. When that data can stay in India, AI moves from pilots into the systems that matter most. Adding in-country inference marks a significant step forward in what we are building here—a growing local team, strong performance in Indian languages, and partnerships that put this technology to work for India.”
“India is at an inflection point in its AI journey, and institutions from banks to government agencies are ready to harness frontier AI at scale,” said Sandeep Dutta, President, AWS India and South Asia. “With Anthropic’s Claude now available for in-country inferencing through Amazon Bedrock, they can deploy it on the same trusted infrastructure they already run their most critical workloads on, unlocking new possibilities in citizen services and digital transformation.”
In February, we officially opened our Bengaluru office—in the months since, our India team has grown significantly, providing applied AI expertise to enterprises, digital natives, and startups building on Claude.
Some of the largest enterprises are deploying Claude across their organizations. Axis Bank leverages Claude to drive engineering excellence and employee productivity. NPCI is building AiNxt, its agentic AI platform, with Claude - bringing secure, governed AI to enterprise scale. IndusInd Bank is building with Claude to create an AI-enabled enterprise knowledge platform that empowers employees with trusted information, augments decision-making, and drives productivity across the organisation. At Godrej Industries,Claude runs across internal app development, business analysis, and everyday knowledge work.
India's IT services leaders are putting Claude into what they deliver for clients worldwide. TCS is equipping 50,000 associates with Claude, taking joint solutions to market for regulated industries and building future-ready AI talent through TCS iON; Infosys is combining Claude with Infosys Topaz™ and setting up a dedicated Anthropic Center of Excellence; Cognizant uses Claude in the systems it builds and runs for clients across manufacturing, life sciences, insurance, and other industries; and LTM & LTTS are embedding Claude and Claude Code across LTM's BlueVerse™ AI Delivery Fabric and LTTS's engineering intelligence platforms. Momentum has picked up across other industries as well.
Digital natives are also embedding Claude in the products customers use every day. Swiggy has made some of its flagship services reachable through MCP. Instamart is now the first quick-commerce service where customers can browse and buy more than 40,000 products by natural-language prompt. Zoho is equipping its customers to leverage Claude models for more intelligent decision-making and better results for their businessesInvideo serves creators across 190 nations, and Claude delivers the precise quality Invideo needs across every language and format.
Startups and product teams are also building with Claude. Rocket lets non-technical teams launch production-grade apps and websites in a fraction of the usual time. Atlan’s native Claude Code plugin lets data teams explore and govern their data estate, semantics and ontology in natural language.Eighteen months ago, Cred started rethinking how software should be built in the age of AI. They adopted Claude code to transform developer productivity and move from AI-assisted coding to AI-native engineering.Zamp.ai serves Fortune 500 enterprises that require reliability and precision from the digital workforce they integrate into essential operations. And Freshworks runs Claude Code across its engineering organization, compressing routine build and review work so teams ship faster.
What customers need is partners who pair deep domain knowledge with advanced technical skill, and who can carry both the last mile. When we launched the Claude Partner Network in March, India had some of the highest registration numbers anywhere in the world. Today, a growing number of Indian partners already sit in our Global Premier, Preferred, and Select tiers.
This week, in Bengaluru we are holding our first Claude Certification for Partners to certify 5,000 people across the network. India has the largest technology talent pool globally, and the partners we build with here will deliver for India and for the world.
Today, we work with organizations tackling some of the toughest problems in India.
Our AI for Science program provides free API credits to researchers across India. Kartik Sunagar’s Evolutionary Venomics Lab at the Indian Institute of Science uses Claude to annotate venom data and find shared targets for antivenoms, tackling a nationwide problem that kills an estimated 58,000 people a year. PopVax, an Indian developer of broadly protective mRNA vaccines and precision cancer medicines, uses Claude as an agent to run its protein-design models, review the results, and refine candidates for its tuberculosis, hepatitis C, Strep A, pancreatic, liver, and gastric cancer programs . And Shravan Hanasoge’s group at the Tata Institute of Fundamental Research is building analysis pipelines for stellar data from the Kepler, TESS, and PLATO missions, using Claude to design the algorithms that extract physical properties of stars from that data and to review the astrophysics literature for promising hypotheses.
Beyond the sciences, organizations are using Claude in work that reaches citizens directly. Pratham—one of the largest education nonprofits worldwide— is piloting Claude in a version of its flagship program, Teaching at the Right Level (TaRL). The AI TaRL engine was built on 170+ hours of facilitator-trainer debrief calls and interviews, plus data from 5,000+ students on the specific mistakes children most commonly make . The program sends government teachers a nightly teaching nudge and provides a chatbot drawing on expert mentors, now piloting in 50 villages in Chhattisgarh. Agami, a nonprofit convening 4,000 justice innovators, is surveying 10,000 Indians on how they already use AI for legal problems, so it can be made accurate and useful for people who struggle to access lawyers and courts . Civis, a civic-tech nonprofit, uses Claude on the documents it drafts and reviews for partners including MeitY, NITI Aayog, and the governments of Maharashtra and Karnataka. And Digital Green, whose FarmerChat answered more than a million farmers in six languages last year, uses Claude to verify the advice it gives.
Adding in-country inference will allow us to deepen this work, opening the door for new deployments where personal data has to stay in India.
Supporting cyber defenders
Attackers are using AI to find and exploit vulnerabilities faster than ever, but defenders who move quickly can find those weaknesses first and patch them, raising the security baseline for everyone. Indian organizations are doing exactly that with Claude.
The Data Security Council of India, a leading cybersecurity body, is turning that work into a collective effort with a Claude-powered AI testing sandbox that gives its members across critical sectors a shared environment to test codebases, digital footprints and supply chains.
Some of the most important code to secure is open source: critical infrastructure that is often maintained by small teams, where vulnerabilities can go undetected for years. AI gives defenders a way to close that gap. MOSIP, the Bengaluru-headquartered open-source platform that underpins national digital ID systems serving more than 200 million people across 30 countries, has made Claude an integral part of how code enters its platform.
The same is true of Digital Public Infrastructure built in India and now spreading around the world. OpenG2P, an open-source platform that helps deliver citizen benefits globally, uses Claude to adapt its platform for each new rollout while strengthening its security.
Six months after signing the New Delhi Frontier AI Impact Commitments, we are making steady progress, including running a company-wide effort to improve Claude’s performance in India’s most widely spoken languages.
Our partnership with Karya is central to this work: in a pilot earlier this year, Karya worked with native speakers and domain experts to evaluate Claude models across seven Indian languages in agriculture use cases, showing how community feedback, combined with domain expertise, can catch linguistic, cultural, and factual issues that automated evaluations miss. We will continue to build on these findings and scale our work to improve the performance of our models across languages, domains and cultural contexts that matter for India.