Anthropic Model Cutoff Exposes India's Sovereign AI Vulnerability

U.S. export restrictions force strategic rethink

India's approach to artificial intelligence development hit a critical roadblock when Anthropic abruptly disabled access to its Fable 5 and Mythos 5 models for foreign nationals, complying with a U.S. government export-control directive. The move exposed a fundamental weakness in New Delhi's strategy of building AI applications on top of foreign foundational models rather than developing its own sovereign technology stack.

"The fact that frontier access can vanish overnight on a foreign government's order is the whole problem," Saket Dandotia, co-founder and chief executive at Onetab.ai, told CNBC. His company, which builds AI applications for enterprises, survived the cutoff only because it had diversified across multiple models. But Dandotia warned that "diversification buys time; it doesn't buy independence."

The incident comes as Indian workers lead global AI adoption, with 41% using the technology nearly daily compared to 26% in China and 19% in the U.S., according to an ADP Research report released last week. That widespread usage now underscores India's vulnerability rather than its strength.

Why it matters

The Anthropic episode demonstrates that India's ambition to become an AI innovation hub remains hostage to foreign technology providers and their governments' policy decisions. Without control over chips, foundational models, and computing infrastructure, the country risks losing its competitive edge in AI development overnight—a strategic liability for the world's fastest-growing large economy.

Critical infrastructure gaps remain

India currently lacks three essential components of a sovereign AI stack: domestic cutting-edge chip production, frontier-scale foundation models comparable to leading U.S. or Chinese offerings, and sufficient data center capacity. Government initiatives including an India semiconductor mission, an AI mission, and tax incentives for hyperscalers are underway but may be insufficient.

The private sector is beginning to respond. Sarvam AI, which focuses on building sovereign AI models, raised $300 million at a $1.5 billion valuation on Monday from investors including HCL Technologies, India's third-largest software services company by market capitalization. Yet Sarvam's flagship model contains just over 100 billion parameters—far short of the several trillion parameters experts say India needs for a non-hallucinating foundational model.

Capital and computing power shortfalls

The fundamental constraint is access to capital and computing resources. Indian startups raised $10.5 billion in funding last year, ranking third globally behind the U.S. and U.K., according to private market intelligence firm Tracxn. However, most funding flowed to enterprise applications, retail, and fintech rather than deep-tech companies working on disruptive technologies.

HCL Tech's investment of 14.27 billion rupees ($151 million) in Sarvam represented less than 10% of what the company paid shareholders in dividends during the financial year ending March 2026—illustrating the relatively modest scale of private investment in sovereign AI infrastructure.

Venture capitalist Mohandas Pai has urged Prime Minister Narendra Modi to launch an expanded AI mission, calling existing government programs "too slow, way too small to make any large impact."

Hardware dependency creates additional risk

Current Indian sovereign AI models rely on Nvidia architecture, creating another potential vulnerability. Neil Shah, vice president of research at Counterpoint Research, warned that if the U.S. restricts access to Blackwell chips as it did with China, India would be left without alternatives.

Sridhar Vembu, co-founder of Indian tech multinational Zoho, emphasized the strategic stakes in a post on X, stating that "Technology is the ultimate weapon" and that India must chart its own path forward. Without substantial government investment to address capital and computing power gaps, efforts to build sovereign AI risk remaining ephemeral, according to Manish Agarwal, co-founder of physical AI data company Humyn Labs.

These details were first reported by CNBC in its "Inside India" newsletter.