Anthropic's suspension of Fable 5 and Mythos 5 access for foreign nationals under a U.S. government directive has reopened a debate India had been postponing: whether the country's AI strategy can keep depending on a small group of U.S. frontier labs. The cutoff applies to all non-U.S. citizens, including Anthropic's own foreign-national employees, and lands in a market the company has publicly called its second-largest after the United States. OpenAI has described India the same way. The directive followed by days a partnership Anthropic announced with Tata Consultancy Services to push enterprise AI adoption across India.
The announcement came late Friday. By Saturday morning, Indian founders, investors, and policy advisors were already publicly arguing about what to do next. The Information reported that the White House is unlikely to extend the same restrictions to other AI providers and is privately attributing the action to Anthropic's handling of alleged jailbreak vulnerabilities — a characterization Anthropic disputes. We covered the underlying Amazon security paper and the resulting Commerce Department order earlier this week.
For India, the dispute about cause matters less than the fact of the suspension. Anthropic and OpenAI have both built out India offices, hired locally, and signed enterprise deals over the past year, betting on the country's developer base to drive adoption. That bet now carries a visible geopolitical risk premium.
Key facts
- 01Anthropic suspended Fable 5 and Mythos 5 access for all foreign nationals, including its own foreign-national employees, under a U.S. government directive.
- 02India is Anthropic's and OpenAI's second-largest market after the U.S., making the cutoff disproportionately impactful there.
- 03Mohandas Pai is calling for a ₹500 billion ($5 billion) annual AI fund and a ₹2 trillion ($21 billion) credit guarantee program for compute and chips.
- 04India's existing IndiaAI Mission is funded at just ₹103.72 billion ($1.2 billion) over five years — roughly a quarter of Pai's proposed annual outlay.
- 05Atomicwork CEO Vijay Rayapati warned that teams without full U.S.-citizen rosters now face a structural competitive disadvantage.
Aakrit Vaish, founder of the Indian AI venture platform Activate, said he woke up Saturday "shocked and confused" and plans to push his portfolio companies to reduce reliance on a handful of frontier providers and lean harder on open-source alternatives. The episode, he argued, strengthens the case for treating sovereign AI as a near-term priority rather than a five-year aspiration.
The competitive risk is sharper for startups with cross-border teams. Atomicwork, which has about 25 employees in the U.S. and most of its product engineering in Bengaluru, sits exactly in the affected zone. Co-founder and CEO Vijay Rayapati told TechCrunch that uneven access to frontier models effectively penalizes any company whose AI team isn't fully U.S.-staffed.
“If your AI team is not made up entirely of U.S. citizens, you are at a competitive disadvantage”— Vijay Rayapati, Co-founder and CEO of Atomicwork
The labor question is already live. This week, U.S. real-estate technology company Opendoor shut its India office less than two years after expanding there, with CEO Kaz Nejatian citing a shift toward smaller AI-native teams closer to U.S. customers. Opendoor did not break out how much of the decision was AI-driven, but the timing has fed a wider argument about whether AI is compressing the economics of offshore engineering talent.
Among India's larger tech voices, the response has been to push for state action. Zoho founder Sridhar Vembu argued on X that "technology is the ultimate weapon" and called on Indian organizations to embrace smaller models, including Indian and Chinese open-source releases. Investor and former Infosys executive Mohandas Pai went further, proposing an annual ₹500 billion (about $5 billion) fund for AI and deep tech alongside a ₹2 trillion (around $21 billion) credit guarantee program covering cloud infrastructure, hardware, and semiconductors.
“We are way behind and need a national mission to get going quickly”— Mohandas Pai, Investor and former Infosys executive
Those numbers would dwarf India's current commitment. The IndiaAI Mission, approved in 2024, allocates ₹103.72 billion (about $1.2 billion) over five years for compute, startup support, and indigenous model development. Pai's proposed annual outlay is roughly four times the five-year mission budget. Whether New Delhi has the appetite to move at that scale is a separate question.
India's foundation-model bench remains thin. Sarvam released open-source models earlier this year, but Krutrim — once positioned as a foundational-model contender — has pivoted toward cloud and AI infrastructure services. Most of India's AI activity sits on top of foreign foundation models. Avataar AI, which we covered earlier this week, launched a video-generation model pitched as a lower-cost alternative to Google's Veo, Kling, Luma, and Runway — a vertical-application play rather than a frontier-model play.
Not everyone thinks capital is the binding constraint. Lightspeed partner Hemant Mohapatra, responding to Pai on X, argued that talent, compute access, and execution matter more than headline fund sizes. He estimated that training a frontier model can cost anywhere from hundreds of millions to several billion dollars depending on the approach, and noted that successful labs have scaled capital requirements alongside adoption rather than front-loading them. There is also the open question of whether the U.S. directive will hold, be narrowed, or be reversed entirely under pressure from Anthropic and its Indian customers.
Prasanto Roy, a New Delhi-based technology policy advisor to multinationals, compared the moment to the lesson many governments drew from Russia's loss of SWIFT access after its invasion of Ukraine. He expects a nationalist backlash in India and called the U.S. decision poorly considered, with consequences well beyond Anthropic. "Even if this is corrected or reversed, the Anthropic episode shows there's no such thing as a geopolitically neutral foreign LLM," Roy said. "American AI models are bound to American geopolitics."
The market consequence is that every enterprise procurement conversation in India now carries an implicit second question alongside price and performance: what happens if Washington pulls the plug. That is a tailwind for open-weight providers — Llama, Mistral, DeepSeek, Sarvam — and for any Indian infrastructure play that can credibly host them at scale. It is a headwind for Anthropic and OpenAI's India revenue lines specifically, and for the broader U.S. frontier-lab pitch that closed-weight APIs are the default substrate for serious AI work. One directive does not redraw the global model market, but it does give every CIO outside the U.S. a defensible reason to diversify, and that decision compounds.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
