Anthropic has disabled access to Fable 5 and Mythos 5 for foreign nationals following a US export control directive, the official explanation was narrow about compliance, national security, and temporary disruption. But in India’s technology circles, the reaction was immediate and critical.
This is not just about two models going dark. It is about the realisation that access to frontier AI and the availability of critical capabilities can change overnight at the discretion of platform owners and governments.
Zoho’s Sridhar Vembu was quick to respond. “Technology is the ultimate weapon. National sovereignty, national security, all of it is now about technology,” he said on X. “Globalisation is dead and Bharat must find her own way.”
His statement, if said a year ago, might have sounded exaggerated. But not anymore.
Within hours, Former Infosys CFO, Mohandas Pai took the argument straight to PM Narendra Modi. He sought a new India AI Mission under the Prime Minister, with Nandan Nilekani as vice chair, backed by serious capital and urgency.
“We are way behind and need a national mission to get going quickly,” Pai said on X, adding that existing government programs are moving too slowly.
As discussed among the industry folks, India may need to allocate ₹50,000 crore annually for deep tech, alongside a larger guarantee framework to crowd in private capital. That is not a policy tweak. That is an attempt to match the scale at which the US and China are already operating.
Rajan Anandan, Managing Director of PeakXV Ventures has also called on PM Modi on X saying, “Why Sovereign AI? This (Fable’s rollback) should put the debate to bed.”
What About the Current Mission?
India does have an AI mission. The 2024 IndiaAI Mission, with an outlay of ₹10,372 crore, was designed to build compute infrastructure, support startups, and expand access. It is a foundational step. But the Anthropic moment has exposed the gap between intent and intensity. In AI, timelines compress fast. A program that looks adequate on paper can look underpowered in practice within months.
As of now, 5 models have been released under the IndiaAI Mission which include models by Sarvam, BharatGen, Tech Mahindra, Gnani.ai, and the latest indigenous AI video model by Avataar. The rest are also in the pipeline.
But what has changed here is not India’s ambition. It is the global context around it.
Speaking of which, Hemant Mohapartra from Lightspeed India, one of the early backers of Sarvam, said that the “sovereign AI is real” moment is here.
“Nation-states will soon start needing citizenship and/or security clearances to work on the next SOTA models the way they do for defense, space, nuclear tech. It is only a matter of time. Talent wars here will be crazy,” he said on X.
For years, the working assumption in much of the ecosystem was that India could plug into global AI platforms while building its own capabilities gradually. That assumption now looks fragile. If access to advanced models can be restricted based on nationality rather than usage, then dependency is no longer just a technical risk. It is a strategic one.
This is also why several people had questioned Nilekani in 2024 about his idea of building India the AI use case capital again. “Our goal should not be to build one more LLM. Let the big boys in the (Silicon) Valley do it, spending billions of dollars,” he had said, while speaking at Meta’s Build with AI summit in Bengaluru in 2024 and reiterated it several times over the last years.
Also read: In 2016, PM Modi Saw an AI Demo. 10 Years Later, Its Creator Says India Still Doesn’t Get AI
But there is also a counterpoint that refuses to go away. Why should the government carry the burden alone?
India’s private sector has built global software businesses, scaled platforms, and delivered cost efficiency at unmatched levels. But frontier AI is a different game.
Anand Ranganathan, Author and Professor at the Special Centre for Molecular Medicine, Jawaharlal Nehru University (JNU), argued that India’s experience in developing cryogenic engine technology offers a lesson for the AI era. Recalling how the United States denied access to crucial LOX/LH₂ cryogenic engine technology, he noted Indian scientists eventually developed the capability indegenously.
“Spine always grows in the back that is against the wall…India can remain sovereign only if it has a sovereign AI,” he said on X.
The Heavy Cost
But frontier AI demands upfront capital, long timelines, and a tolerance for failure that most companies have not had to internalise at this scale. The uncomfortable truth is that while leaders are asking for a national mission, the ecosystem itself is still calibrating how much risk it is willing to take.
Mohapatra from Lightspeed explained in another post that building a frontier model is not just about ambition, it is about sustained capital and capability.
He said that training a GPT-class 1 trillion parameter model from scratch, including failed runs, data acquisition, cleaning, RLHF, and post-training, can require about $250 million in compute on an aggressive 3-4 month timeline, and roughly $500-600 million all-in for a dense model.
But pushing toward state-of-the-art technology, it raises the bar sharply. “If you want SOTA,” he notes, budgets can climb to $2-4 billion once you factor in test-time compute, post-training reinforcement learning, synthetic data generation, and, most critically, talent. However, the real challenge begins with serving the model at scale and ensuring enough usage to justify the investment.
Mohapatra’s larger point is not about raising $50-60 billion upfront. Companies like Mistral, OpenAI, and Anthropic have all scaled capital over time as adoption grew. The early bottleneck, he argues, is far more basic and far more difficult: access to top-tier talent and enough GPUs “at that scale where you can do interesting things.”
The Anthropic decision has, in that sense, forced a rethinking. The state cannot move at a measured pace if the technology is moving exponentially. And the private sector cannot wait for policy cover before committing capital and talent.
India’s push for sovereign AI is rooted in a familiar concern. Access to critical technologies can change, and AI is increasingly being viewed as one of those strategic capabilities.
Your reaction