India revenue run rate has doubled over the last four months: Anthropic CEO Dario Amodei

Anthropic India’s revenue run rate has doubled over the last four months, said CEO Dario Amodei, adding that adoption of its Claude Code may have grown even faster among developers. A company’s revenue run rate projects its future annual revenue based on the current quarterly or monthly revenue.

"We can run experiments here involving hundreds of millions of people. That scale allows entrepreneurs and builders to learn incredibly fast and fail fast. This is something that’s much harder to do in smaller markets.

"It’s incredible to see the pace at which things are happening. It mirrors the broader explosion in Claude models and coding models globally, but the intensity in India feels even more extreme than what we’re seeing elsewhere," he said while speaking at the company's Builder Summit in Bengaluru.


In a conversation with Irina Ghose, managing director, Anthropic India, Amodei discussed the India opportunity, the growth of enterprise AI globally, and coexisting with AI systems.

On what stands out in India, Amodei said it's the "ability to build software with AI and to build on top of AI, developing new applications and systems." Citing an example, he said, "I heard recently that the ministry of statistics is building a server to query economic data. Government bodies elsewhere typically don’t move this fast or engage this deeply with emerging technology."


When Ghose asked about how Indian developers should use Anthropic's tools, Amodei said he's already seeing a number of exciting and unique applications coming out of India. "One major area is around building systems that interoperate across (Indian) languages and make translation and multilingual access easier." He added that Claude is working hard to support the long tail of languages, where India’s linguistic diversity opens up opportunities.

AI for the future

Using an ice hockey metaphor, Amodei said developers or startups should “skate to where the puck is going,” and added that, "You shouldn’t build for where AI models are today, you should build for where they’ll be in a year or two."

Amodei explained that since large enterprises take time to deploy and test systems, they need to imagine where the technology is headed: ``Don’t just automate step four of a 10-step workflow. Try to automate the whole workflow, even if the model isn’t fully capable yet,’’ Amodei said, addressing about 250 developers on Monday.

For the Anthropic CEO, the most exciting area is the usage of AI models in medicine and health. "I strongly encourage builders to explore where AI intersects with these fields. Pure software apps are easy to replicate, but applications tied to the physical world — biology, medicine, etc., — end up creating real moats." He explained that these fields require deep domain knowledge, regulatory navigation, and long-term effort. "That’s hard, but it leads to durable businesses and meaningful impact."

Amodei said the ecosystem is moving towards true end-to-end workflows, where humans act as supervisors and can accelerate productivity by 10 or even 100X.

Also Read: Anthropic to expand India operations, invest behind Indic language AI models