Anthropic’s recent decision to halt access to its latest AI models in response to a U.S. government request has sparked renewed discussions within the global tech sector. In India, this action has reignited an ongoing debate about the country’s reliance on AI technologies developed elsewhere, particularly given its standing as a significant player in the AI landscape.
The announcement was made late on a Friday, revealing that Anthropic had received a directive from the U.S. government to restrict access to its recently introduced Fable 5 and Mythos 5 models for foreign nationals, including its employees from abroad. This decision came shortly after the company formed a partnership with Tata Consultancy Services, one of India’s leading IT service firms, aimed at promoting enterprise-level AI adoption in the country, highlighting the intricate ties between India’s AI ambitions and U.S.-based technology.
While the full implications of this action are yet to be understood, some reports indicate that initial security concerns were raised by Amazon’s CEO, Andy Jassy, leading to this government directive. The White House reportedly does not plan to impose similar restrictions on other AI companies and is attributing the situation to Anthropic’s management of potential vulnerabilities. Anthropic has contested the government’s portrayal, suggesting that this action was unwarranted.
Nonetheless, the situation has spurred debate among Indian entrepreneurs, investors, and policy experts regarding whether India should hasten its efforts to develop its own AI capabilities, invest more in open-source alternatives, or continue depending on a limited number of U.S. AI providers. For many, the incident serves as a wake-up call regarding technological dependency, while for others, it underscores the reality that access to vital AI systems can be influenced by geopolitical factors outside India’s purview.
India is increasingly recognized as a crucial market for leading AI firms, with both Anthropic and OpenAI ranking it as their second-largest market after the United States. This highlights India’s growing role in the global AI competition, prompting these companies to establish local offices and increase hiring and partnerships in the country, banking on the large pool of developers, startups, and businesses to drive the adoption of their technologies.
For many stakeholders in India’s tech landscape, Anthropic’s announcement extends beyond just one company’s operations. It raises fundamental questions about India’s long-term AI strategy and the viability of continuing to depend on a limited number of foreign AI technologies.
“It completely changes things,” stated Aakrit Vaish, founder of the AI venture platform Activate, in reference to Anthropic’s decision. He noted that this situation underscores the need for India to prioritize domestic AI development. Vaish expressed a shock over the news and indicated a shift toward open-source models among startups, encouraging entities in his portfolio to lessen their reliance on a few leading AI providers.
For some founders, there are broader concerns about the competitive implications of restricted access to frontier AI technologies. Vijay Rayapati, CEO and co-founder of Atomicwork, shared that the situation exposes risks for startups with geographically diverse teams, where restrictions might become a geopolitical issue affecting advanced AI system accessibility.
Atomicwork employs around 25 staff in the United States, while a substantial part of its engineering team operates from Bengaluru, India.
“If your AI team is not composed solely of U.S. citizens, you are inherently at a disadvantage,” remarked Rayapati, emphasizing that limited access to cutting-edge AI models could unfairly benefit certain companies over others.
The timing of these developments coincides with ongoing debates in India’s tech sector about AI’s potential to alter global labor economics. A notable incident involved Opendoor, a U.S. property technology firm, closing its India office within two years of its establishment, with CEO Kaz Nejatian attributing the shift to an operational realignment toward U.S. clients and a preference for smaller, AI-focused teams.
Although Opendoor did not clarify how much AI-related efficiencies influenced this decision, it contributes to a wider discussion regarding how AI advancements might impact the landscape of global technology roles and India’s position as a hub for engineering talent.
Looking Beyond Anthropic
The situation with Anthropic has also sparked a broader discussion within India’s tech leadership concerning reliance on foreign AI infrastructure.
Sridhar Vembu, founder of SaaS company Zoho, commented that this incident illustrates how technology can become a significant geopolitical tool, advocating for a greater embrace of smaller and open-source models within Indian organizations.
“What steps can our government take now? We should ensure that organizations in India adopt smaller models, including open-source options from Indian and Chinese sources,” Vembu expressed on X.
Investor and former Infosys executive Mohandas Pai responded to Vembu’s comments by pointing out the urgent need for a more robust national AI strategy, calling for increased investments in AI technologies, computing infrastructure, and deep tech innovations.
“We are lagging significantly and require a national mission to expedite progress,” Pai contended, proposing the establishment of a ₹500 billion (approximately $5 billion) annual fund for AI and deep tech, alongside a ₹2 trillion (around $21 billion) guarantee program to bolster cloud infrastructure and hardware development.
Such proposals would far exceed the size of India’s current AI-focused initiatives. In 2024, the IndiaAI Mission was approved with a budget of ₹103.72 billion (around $1.2 billion) allocated over five years to enhance computational capabilities, support startups, and foster indigenous AI advancements.
Despite a rising interest in AI and governmental efforts to cultivate domestic skills, India remains relatively small concerning frontier AI model development. Only a few startups are attempting to create foundational AI models, with Sarvam recently introducing open-source models. Conversely, another notable startup, Krutrim, has shifted its focus from foundational development to cloud and AI infrastructure services.
Most of India’s AI ecosystem appears to concentrate on building applications and specialized models on existing foundational technologies. For instance, Avataar AI recently launched a video generation model aimed at providing a cost-effective alternative to competitors like Google’s Veo and others.
There is not a consensus that the primary hurdle is a lack of funding. Responding to Pai’s comments, Lightspeed partner Hemant Mohapatra argued that the main obstacles in establishing competitive AI enterprises are talent shortages, access to computing resources, and execution capabilities, rather than just capital availability.
Mohapatra estimated the costs for training a frontier AI model could range between hundreds of millions to several billion dollars depending on the method used, yet noted that successful AI companies often scale their funding needs as adoption progresses.
For some experts in policy, the ramifications of this situation extend beyond AI startups and model developers.
Prasanto Roy, a technology policy analyst in New Delhi, noted that the incident will likely heighten concerns among Indian policymakers about strategic autonomy, drawing parallels to the repercussions many nations faced following Russia’s exclusion from the global financial system after its actions in Ukraine.
He suggested that this development could provoke a significant nationalist response within India and characterized it as a misguided decision by the U.S. government, with effects likely reaching far beyond Anthropic.
“Even if this situation is corrected or rescinded, the Anthropic incident demonstrates there is no such thing as a geopolitically neutral foreign AI model,” Roy emphasized. “American AI tools will inherently align with U.S. geopolitical interests.”
