India's AI strategy is crumbling under the weight of geopolitical restrictions, forcing the country to confront a hard truth: building applications on top of foreign AI models leaves startups vulnerable to overnight shutdowns. When Anthropic disabled access to its new Fable 5 and Mythos 5 models for foreign nationals last week, complying with U.S. export controls, it exposed the fragility of India's entire approach to artificial intelligence.

What Happened When Anthropic Cut Off Access?

The restriction hit Indian startups immediately. Saket Dandotia, co-founder and chief executive at Onetab.ai, a company building AI applications for enterprises, found his business model suddenly at risk. "The fact that frontier access can vanish overnight on a foreign government's order is the whole problem," he told CNBC. While Dandotia had diversified across multiple AI models to protect himself, he acknowledged a hard reality: "diversification buys time; it doesn't buy independence".

This single moment crystallized what Indian policymakers and tech leaders have been reluctant to admit. India's AI strategy, which relied on leveraging the country's vast information technology talent to build applications using foreign foundational models, had a fatal flaw. Without control over the underlying AI models themselves, Indian companies remain perpetually at the mercy of U.S. export controls and geopolitical decisions made in Washington.

How Is India Currently Using AI, and Why Is That a Problem?

The scale of India's AI adoption is impressive on the surface. An ADP Research report found that 41% of Indian workers use AI nearly every day, higher than 26% in China and 19% in the U.S. . But this statistic masks a troubling dependency. That high adoption rate reflects how deeply Indian workers and companies rely on foreign technology, not Indian innovation.

India lacks three critical components of a sovereign AI stack:

• Cutting-Edge Chip Manufacturing: India does not yet produce advanced semiconductors domestically, making it dependent on imports from countries like Taiwan and the United States.
• Frontier-Scale Foundation Models: India has no AI model comparable to leading U.S. or Chinese systems; Sarvam AI's flagship model, for example, has just over 100 billion parameters, far smaller than frontier models with trillions of parameters.
• Data Center Capacity: India's data center infrastructure lags considerably behind the U.S. and China, limiting the computing power available for training large AI systems.

The government is attempting to address these gaps through an India semiconductor mission, an AI mission, and tax breaks for global hyperscalers setting up data centers in the country. But experts warn these efforts may be moving too slowly.

What Are the Biggest Barriers to Building Sovereign AI in India?

The private sector is beginning to recognize the urgency. On Monday, Sarvam AI, which is building sovereign AI models, raised $300 million at a $1.5 billion valuation from investors including HCL Technologies, India's third-largest software services company by market cap. Yet even this significant investment reveals the capital shortage India faces.

HCL Technologies' investment of 14.27 billion rupees, approximately $151 million, was less than 10% of what the company paid out to shareholders as dividends in the financial year ending March 2026. This illustrates how venture capital in India remains conservative when betting on deep-tech companies, the firms working on cutting-edge disruptive technologies that sovereign AI requires.

"The biggest challenge for India is access to computing power and a lack of deep-tech investment capital," explained Manish Agarwal, co-founder of Humyn Labs, a physical AI data company.

Manish Agarwal, Co-founder of Humyn Labs

Indian startups raised $10.5 billion in funding last year, the third highest in the world after the U.S. and U.K., according to Tracxn. However, most of these funds went to startups in enterprise applications, retail, and fintech sectors, not deep-tech companies. Venture capitalists in India invest far smaller amounts in deep-tech startups compared with the billions being invested overseas.

Building a foundational AI model that does not hallucinate, or produce false information, requires a few trillion parameters for a country the size of India, experts said. This demands enormous capital and computing power. Sarvam's flagship model, with just over 100 billion parameters, falls far short of what would be needed to compete globally.

What Risks Does India Face If It Relies on Nvidia Chips?

There is another vulnerability lurking beneath India's sovereign AI ambitions. Currently, the AI models being developed in India use Nvidia architecture, the dominant chip design for training large language models. But if the U.S. restricts access to Nvidia's Blackwell chips as it did with China, India would be left helpless, according to Neil Shah, vice president of research at Counterpoint Research.

This risk is not theoretical. Influential voices in the Indian tech industry are already sounding the alarm. Sridhar Vembu, co-founder of Indian tech multinational Zoho, posted on X that "Technology is the ultimate weapon," and that India must find its "own way ahead". But without capital and sovereign computing power, that path remains blocked.

Influential

What Would It Take for India to Build True Sovereign AI?

Prominent venture capitalist and angel investor Mohandas Pai has urged Prime Minister Narendra Modi to launch a major AI mission, calling existing government programs "too slow, way too small to make any large impact". Without a strong government drive to address the capital and computing power shortages, the conversation around building sovereign AI risks becoming ephemeral, or temporary and fleeting, according to experts.

The stakes are high. India's strength lies in its strong domestic market at both consumer and enterprise levels, and its deep pool of tech talent. But without the capital and infrastructure to build sovereign AI, that talent will continue to be channeled into building applications on top of foreign models, leaving India perpetually vulnerable to geopolitical restrictions.

The Anthropic restriction was a wake-up call. Whether India heeds it remains to be seen.