India’s strategic reliance on foreign foundational AI models is under scrutiny after Anthropic restricted access to its new models, exposing a critical gap in the nation’s sovereign AI capabilities. Despite ambitious plans to leverage its IT talent for AI applications, experts warn that India’s current initiatives in chip development, foundation models, and data centers are “too slow, way too small” to achieve true independence, highlighting severe shortages in capital and computing power. This reliance on external tech leaves Indian startups vulnerable to foreign directives, fueling urgent calls for greater domestic investment and a more robust government-led AI mission.
Hello and welcome to the latest edition of "Inside India," bringing you critical insights from the world's rapidly expanding major economy. This week, we delve into a pressing issue that has forced New Delhi to urgently reconsider its artificial intelligence strategy: the recent restrictions imposed by the U.S. government on certain AI offerings, exemplified by Anthropic's latest export curbs.
For a long time, India's vision for AI was clear: become a global innovation hub by leveraging its immense information technology talent to develop cutting-edge applications, all built upon foundational models from overseas. This strategy, however, has proven fragile.
A Shifting Landscape: Anthropic's Curbs Expose Vulnerability
The vulnerability of India's AI ambitions was starkly revealed last week when Anthropic, a prominent AI developer, abruptly disabled access to its new models—Fable 5 and Mythos 5—for foreign nationals. This move, in compliance with a U.S. export-control directive, sent ripples through India's tech ecosystem.
"The fact that frontier access can vanish overnight on a foreign government's order is the whole problem," emphasized Saket Dandotia, co-founder and CEO of Onetab.ai, in an interview with CNBC. Dandotia, whose business relies on creating AI applications for enterprises, recounted that his operations could have been severely disrupted had he not already diversified across multiple AI models. Yet, he cautions, "diversification buys time; it doesn't buy independence." He asserts the critical need for India to develop sovereign AI to safeguard its startups from the whims of foreign policy.
While an ADP Research report highlights India's high AI adoption rate, with 41% of Indian workers using AI almost daily (compared to 26% in China and 19% in the U.S.), this statistic inadvertently underscores the nation's deep reliance on foreign technological infrastructure.
The Sovereign AI Gap: Challenges and "Too Slow, Too Small" Efforts
India currently faces significant hurdles in building a comprehensive sovereign AI stack. Domestically, it lacks the capacity to produce cutting-edge chips and is yet to develop a frontier-scale foundation model comparable to those from the U.S. or China. Furthermore, its data center capacity, though growing, significantly trails these global leaders.
The Indian government has initiated efforts on multiple fronts, including an India semiconductor mission, an AI mission, and tax incentives for global hyperscalers establishing data centers. The private sector is also stepping up; notably, Sarvam AI, an Indian firm focused on sovereign AI models, recently secured $300 million in funding at a $1.5 billion valuation, with India's third-largest software services company by market cap, HCL Technologies, among its investors.
Despite these initiatives, industry experts remain largely unconvinced. Manish Agarwal, co-founder of Humyn Labs, a physical AI data company, suggests that current efforts might be "too little, and likely too late." He points to a critical dual challenge: access to substantial computing power and a scarcity of deep-tech investment capital.
India boasts a vibrant domestic market and a deep pool of tech talent, but it struggles to attract the colossal capital available to sovereign AI ventures in the U.S. and China. Last year, Indian startups attracted $10.5 billion in funding, placing it third globally after the U.S. and U.K. However, the lion's share of this investment flowed into enterprise applications, retail, and fintech, largely bypassing deep-tech companies—firms dedicated to pioneering disruptive technologies.
Venture capitalists in India tend to be more conservative when investing in deep-tech ideas. For instance, HCL Tech's investment of 14.27 billion rupees ($151 million) in Sarvam represented less than 10% of its shareholder dividends for the financial year ending March 2026. This disparity highlights the urgent need for more significant capital infusion.
India's Prime Minister Narendra Modi (C) takes a group photo with AI company leaders including OpenAI CEO Sam Altman (2nd R), Anthropic CEO Dario Amodei (R), Google CEO Sundar Pichai (2nd L), and Meta Chief AI Officer Alexandr Wang (L), at the AI Impact Summit in New Delhi on February 19, 2026.
Ludovic Marin | Afp | Getty Images
Consequently, calls for the government to invest more heavily in developing sovereign AI are intensifying. Mohandas Pai, a prominent venture capitalist, has openly urged Prime Minister Narendra Modi to launch a dedicated AI mission, dismissing existing government programs as "too slow, way too small to make any large impact."
Experts estimate that building a robust, foundational AI model for a country of India's scale, one that can avoid "hallucinations," would require trillions of parameters, demanding immense capital and computing power. Sarvam's flagship model, for comparison, currently has just over 100 billion parameters.
Another critical concern is India's reliance on Nvidia architecture for its current sovereign AI models. Neil Shah, vice president of research at Counterpoint Research, warns that if the U.S. extends restrictions on advanced chips like Blackwell to India, similar to its actions against China, the nation could find itself "helpless."
This sentiment is echoed by other influential figures in the Indian tech industry. Sridhar Vembu, co-founder of Indian tech multinational Zoho and who founded Arattai, emphasized in a post on X that "Technology is the ultimate weapon," and India must "find its own way ahead." Yet, achieving this independence hinges on securing both substantial capital and sovereign computing power—resources that India currently lacks significantly.
Without a powerful government-led initiative to address these fundamental deficiencies, the burgeoning conversation around building sovereign AI in India risks becoming merely "ephemeral," cautions Agarwal of Humyn Labs.
In Other News: Quick Takes from India
- Messaging platform Telegram voiced strong disapproval over India's temporary ban on its app, citing concerns over exam fraud. Telegram argued the move unjustly punishes its 150 million ordinary users rather than targeting those responsible for leaking test papers.
- The U.S. has emerged as India's leading supplier of liquefied natural gas (LNG) and liquefied petroleum gas (LPG) in May. This shift occurred as shipments from traditional Gulf suppliers declined due to disruptions in traffic through the Strait of Hormuz, likely related to the Iran war.
Coming Up:
- June 19: Reliance Industries annual general meeting.
- June 23: HSBC Composite Flash PMI for June.
