India's AI Ambition Hits a U.S. Export Control Wall

India's strategy to become an AI powerhouse by building applications on top of foreign models looked solid - until last week. Anthropic blocked access to its latest models for foreign users, complying with a U.S. export-control order. For startups like Onetab.ai, which had bet on those models, the rug was pulled out from under them. CEO Saket Dandotia told CNBC he had diversified across multiple models, but added, "Diversification buys time; it doesn't buy independence."

The incident has forced New Delhi to rethink its approach. Indian workers use AI at a higher rate than their peers in China or the U.S., but that adoption masks a deep dependency on foreign technology. The country lacks domestic chip fabrication, a frontier-scale foundation model, and sufficient data center capacity.

Government initiatives - a semiconductor mission, an AI mission, and tax breaks for hyperscalers - are underway. The private sector is stirring, too: Sarvam AI recently raised $300 million at a $1.5 billion valuation, with backing from HCL Technologies. But industry experts say these efforts are too small and too slow. India's deep-tech startups struggle for capital; venture funding flows mostly to enterprise apps, retail, and fintech, not to foundational AI.

Mohandas Pai, a prominent venture capitalist, has urged Prime Minister Modi to launch a more aggressive AI mission, calling current programs "too slow, way too small to make any large impact." Without a stronger government push, experts warn, India's sovereign AI ambitions risk fading into ephemerality.

Source: CNBC