The shutdown of Anthropic's model has provided a significant boost to India's sovereign AI initiative, giving it a compelling argument.

      India is currently engaged in discussions about sovereign AI following the US government's directive for Anthropic to shut down Fable 5, which has resulted in proposals for a $5 billion fund and an emphasis on embracing open-source models.

      On June 12, 2023, the US government mandated Anthropic to discontinue Fable 5 and Mythos 5, targeting export control to prevent foreign nationals from accessing the nation’s advanced AI technologies. This move served as a stark reminder to India, Anthropic’s second-largest market, of the implications when AI infrastructure is tied to external political agendas.

      As a consequence, Indian developers and businesses were cut off from access to Claude’s most advanced models almost immediately. Since October 2025, India’s revenue from Claude had doubled, and just a day prior, Tata Consultancy Services had announced a collaboration to train 50,000 employees on Claude and establish a dedicated Anthropic business unit, leaving that agreement in uncertainty.

      This timing has escalated an already existing debate about AI sovereignty into a pivotal strategic discussion. What once appeared to be ambitious proposals now seem urgent.

      Mohandas Pai, the former CFO of Infosys and a leading tech investor in India, has proposed a ₹50,000 crore (approximately $5 billion) annual sovereign AI fund. Additionally, he suggests a ₹2 lakh crore (around $21 billion) credit guarantee to support cloud infrastructure, hardware acquisition, and semiconductor development. These amounts significantly surpass the government's current commitments.

      India had recently approved its IndiaAI Mission in March 2024 with a budget of ₹10,372 crore (about $1.25 billion), which has deployed around 38,000 GPUs to date. Pai’s proposal would increase annual expenditures fourfold and introduce a credit backstop substantially larger in scale.

      Sridhar Vembu, CEO of Zoho, advocates for India to adopt smaller and open-source models, including those from China, rather than rely on American advanced systems that can be easily disabled by executive decision. “Technology is the ultimate weapon,” Vembu remarked, asserting that the globalization era is over and that India must carve its own path forward.

      This argument gains weight as the suspension highlights the very vulnerabilities Vembu points out. Reports indicate that the action to limit access was spurred by Amazon’s CEO informing Treasury Secretary Scott Bessent that researchers had leveraged Fable 5 to gather information for potential cyberattacks. Although Anthropic deemed the response excessive, compliance was swift and comprehensive.

      Prasanto Roy, a policy expert, succinctly stated: “American AI models are bound to American geopolitics.” For Indian businesses reliant on Claude, the key lesson was that access to advanced AI can be withdrawn at any time, without warning, consultation, or consideration of the disrupted commercial relationships.

      The Indian startup sector is already evolving in response. Sarvam, a Bengaluru-based AI firm, introduced open-source models with 30 billion and 105 billion parameters at the India AI Impact Summit in 2026. Krutrim, started by Ola’s Bhavish Aggarwal, shifted from foundational model development to offering cloud and AI infrastructure services, generating ₹3 billion in revenue for fiscal 2026.

      Neither company currently matches the capabilities of Fable 5 or Mythos 5. However, the case for sovereign AI is less about immediate parity with frontier performance and more about ensuring stability when Washington makes unilateral decisions regarding model access.

      Aakrit Vaish, founder of the AI startup Activate, remarked that the suspension has significantly impacted the sovereign AI discussion. Meanwhile, Vijay Rayapati, CEO of Atomicwork, expressed concerns about the implications for Indian firms that operate with international teams reliant on American AI services. Should the US limit model access to enforce export policies, any nation dependent on American AI risks disruption from a single policy change.

      Not all share the view that India should develop its own advanced models. Hemant Mohapatra, a partner at Lightspeed Venture Partners, contended that access to talent and computing resources is more critical than capital for achieving competitive AI. While India has a substantial engineering workforce, there is a notable gap in computing resources, which requires either significant domestic investment or ongoing access to foreign cloud services.

      Anthropic had established an office in Bengaluru as part of its expansion, and the TCS partnership was intended to bolster its enterprise strategy within India. The sustainability of these plans hinges on Anthropic's ability to quickly restore access and whether Indian enterprises can retain trust in a provider whose most advanced models can disappear without warning.

      The overarching trend is clear. The US has spent the last four years strengthening controls over AI technology, from chip export restrictions to direct interventions at the model level. Each intensification nudges more countries towards the realization that reliance on American AI infrastructure poses political risks. With its 1.4 billion population and rapidly expanding tech sector, India is now questioning the feasibility of that risk and what it would entail to mitigate it.

      Additionally, the layoffs at Opendoor