Anthropic sets up shop in Bengaluru

If you want to see where AI coding tools are being pushed to their absolute limits, look at India.

Anthropic has officially cut the ribbon on its new office in Bengaluru, marking its second outpost in Asia following Tokyo. Helmed by tech veteran Irina Ghose as Managing Director, the expansion comes alongside an aggressive slate of partnerships spanning enterprise IT, public infrastructure, and grassroots nonprofits.

India is already Claude.ai's second-largest market globally. Anthropic revealed that its run-rate revenue in the country has doubled since it first signaled expansion plans back in October 2025. Even more telling is how locals use it: nearly half of all Claude traffic in India consists of technical computer and math tasks, from modernizing legacy software to pushing code straight to production.

Enterprise adoption and national-scale MCP servers

India's massive IT sector is leaning into Claude hard. Air India is deploying Claude Code across operations to spin up custom software faster. Fintech darling CRED reported that Claude Code helped them double their feature delivery speed while improving test coverage by 10%. Meanwhile, IT services giant Cognizant is rolling Claude out to a massive 350,000 employees globally to overhaul client systems.

Startup traction is moving at a similar breakneck pace. Emergent, a natural-language app builder running entirely on Claude, hit $25 million in annual recurring revenue and two million users in under five months.

Perhaps the most fascinating development is the government footprint. Following Anthropic's donation of the Model Context Protocol (MCP) to the Linux Foundation, India's Ministry of Statistics and Programme Implementation just deployed the country's first official government MCP server, letting AI systems query national data directly. In the private sector, delivery giant Swiggy is using MCP to let users order groceries and book tables right inside Claude.

Tackling 50 million pending court cases

Expanding in India also means cracking the language barrier. Six months ago, Anthropic kicked off an initiative to build high-quality datasets across 10 of India's major languages, including Hindi, Bengali, Tamil, and Urdu. The company is now partnering with Karya and the Collective Intelligence Project to publish open evaluations for local legal and agricultural contexts.

That groundwork is already powering public-interest deployments. India has roughly 50 million pending court cases, where tracking updates traditionally involves navigating layers of paper files and middlemen. Legal startup Adalat AI just launched a national WhatsApp helpline powered by Claude, letting citizens track status updates and parse legal jargon in their native tongue.

On the education side, nonprofit Pratham has deployed Claude-powered Anytime Testing Machines across 20 schools, aiming to reach 100 schools by late 2026 to help underserved students and women returning to education.

Why it matters

We think Anthropic's Bengaluru launch represents one of the smartest enterprise plays in AI right now. While rivals are busy battling over consumer chatbots and Western SaaS seats, Anthropic is anchoring itself in the world's engineering back office. India's IT services ecosystem is the engine room of global enterprise software, making it the ultimate proving ground for Claude Code.

The strategic use of MCP is also paying off. By getting MCP adopted by government ministries and everyday consumer giants like Swiggy, Anthropic is embedding Claude into the country's digital plumbing rather than treating it like a standalone web interface.

Still, running high-stakes AI in India comes with severe risks. Hallucinations in casual coding are annoying; hallucinations in legal case tracking or rural agricultural advice can derail lives. Navigating these edge cases across a dozen languages will be the real test of whether Anthropic's safety-first philosophy holds up at population scale.