Anthropic curbs are fueling the AI ​​debate in India

Hello, this is Priyanka Salve, writing to you from Singapore.

Welcome to the latest edition of the “Inside India“ — your destination for news and developments from the world’s fastest growing economies.

While the US and China were racing to develop a host of artificial intelligence, India was confident of making a mark by building an AI application layer on top of external base models. But now New Delhi is being forced to rethink its strategy as Washington moves to limit certain AI contributions.

Any ideas for today’s newsletter? Share them and the group.

India’s strategy for artificial intelligence was simple: become an AI innovation center by using its vast information technology talent to develop applications using external models.

But the fragility of that ambition was revealed last week when Anthropic disabled access to its new models – the Fable 5 and Mythos 5 – to foreigners, complying with an export control order from the US government.

“The fact that access to a border can disappear overnight because of a foreign government order is a whole problem,” Saket Dandotia, founder and CEO of Onetab.ai, told CNBC.

Anthropic’s suspension of access could have broken Dandotia’s business of making AI applications for businesses if it had not been separated into multiple models.

But “diversification buys time; it doesn’t buy independence,” Dandotia said, adding that India needs AI to govern, so that startups like his don’t “lose their edge” with a single directive from a foreign government.

An ADP Research report released on Thursday found that 41% of Indian workers use AI on an almost daily basis, higher than 26% in China and 19% in the US.

India does not yet produce high-end chips domestically, nor does it have a base-scale model on par with the flagship models in the US or China. While growing rapidly, its data center capacity also lags behind the US and China.

Government efforts continue on all three fronts, through the India semiconductor mission, the AI ​​mission, and tax breaks for global hyperscalers setting up data centers in the country.

The private sector is also starting to see the need to invest in domestic AI. On Monday, India’s Sarvam AI, which builds autonomous AI models, has raised $300 million in a $1.5 billion valuation from an investor group that includes India’s third largest software company by market value, HCL Technologies.

But the efforts may be too little, and possibly too late, say industry experts, adding that India’s biggest challenge is access to computing power and a lack of capital to invest in deep technology.

India’s strength is its strong domestic market, both for consumers and businesses, and its deep tech talent, but it lacks the capital available to private AI companies in the US and China, said Manish Agarwal, founder of Humyn Labs, an AI data company.

Last year, Indian startups raised $10.5 billion, the third highest in the world after the US and the UK, according to a report by independent market intelligence firm Tracxn in December. However, most of these funds have gone to start-up business applications, retail, and fintech sectors, not deep-rooted companies. – Companies operating on disruptive old technologies.

Investments made by Indian venture capitalists in deep-tech startups are relatively small compared to the billions invested overseas, experts told CNBC, adding that private investors are more cautious when betting on deep-tech ideas.

For example, HCL Tech’s investment of 14.27 billion rupees ($151 million) in Sarvam was less than 10% of what it paid out to shareholders as dividends in the fiscal year ending March 2026.

Therefore, calls are growing within the country, asking the government to invest more in developing autonomous AI.

Mohandas Pai, a prominent venture capitalist and angel investor, urged Prime Minister Narendra Modi to jumpstart AI, calling existing government programs “too slow, too small to make any significant impact.”

In a country the size of India, building a basic non-emotional AI model would require a few billions of dollars, say experts, adding that this requires a lot of money and computing power. India’s flagship Sarvam model, for example, has more than 100 billion parameters.

Currently, independent AI models being developed in India use Nvidia architecture, but if the US restricts access to Blackwell chips as it has with China, it will leave India with nothing, Neil Shah, vice president of research at Counterpoint Research, told CNBC.

This is a risk voiced by some influential voices in the Indian tech industry.

Sridhar Vembu, the founder of Indian tech multinational Zoho and who founded Arattai to compete with WhatsApp in India, said in a post on X that “Technology is the ultimate weapon,” and that India must find its own “way forward.”

But to do that, the country needs money and massive computing power, both of which are sorely lacking.

Without a strong government campaign to solve those problems, the conversation about building autonomous AI is at risk of “indefinite life,” says Agarwal of Humyn Labs.

Telegram criticizes India for temporarily shutting down the app due to exam cheating issues
Messaging platform Telegram criticized the Indian government’s move to temporarily block access to the app to prevent exam cheating, saying it penalizes 150 million regular users rather than those responsible for leaking test papers.

The US is becoming India’s leading gas supplier, as the war with Iran cuts it off from the Gulf
The United States has emerged as the top supplier of liquefied natural gas and petroleum gas to India in May, as shipments from the Gulf countries fell due to traffic disruptions in the Strait of Hormuz.

It’s coming

June 19: Annual meeting of Reliance Industries.

June 23: HSBC Composite Flash PMI for June.