US export controls have halted foreign access to advanced AI models, prompting Indian venture capital and technology leaders to accelerate domestic sovereign AI infrastructure development. Relying on overseas foundation model providers now exposes enterprises and governments to acute geopolitical risk.
Export controls trigger sovereign AI push
The US government issued an export control directive forcing Anthropic to suspend access to its Fable 5 and Mythos 5 models for all foreign nationals, including those based outside the US. This abrupt shutdown has reinforced the message that relying on a single overseas provider is an unsustainable long-term strategy for Indian startups. "If this is not a wake-up call for the country, I don't know what is," said Aakrit Vaish, cofounder of AI-focused early stage venture capital firm Activate.
Vaish, who previously advised the government on the IndiaAI Mission, argued that policymakers must prioritize sovereign AI labs. He emphasized that India can experiment with new architectures and build smaller models designed specifically for local languages and use cases, rather than chasing the exact capital-intensive approach of frontier labs.
Building indigenous infrastructure
Industry leaders argue India must build its own AI capabilities, drawing parallels to the success of digital public infrastructure like UPI and Aadhaar. Currently, India's sovereign AI push hinges on Sarvam AI, a Bengaluru-based startup selected under the IndiaAI Mission to build indigenous foundation models.
"For AI users, it is clear that you should not confuse access with ownership, or adoption itself as an advantage," said Pratyush Kumar, cofounder of Sarvam AI. Sarvam AI has raised $41 million to date, a fraction of the approximately $350 billion raised collectively by US frontier labs Anthropic and OpenAI. Reports indicate Sarvam is in talks to raise an additional $250 million to $300 million at a valuation of $1.2 billion to $1.5 billion.
To construct these smaller, localized systems, engineers and researchers are increasingly focusing on AI for IT & Development rather than solely depending on massive foreign architectures. This shift requires teams to develop internal capabilities for model training, deployment, and maintenance.
Operational risks for global founders
Over the past two years, more than 100 Indian founders relocated to the US to build AI companies, and these curbs now threaten their survival. "My team and I were using it 24/7 and the moment it was shut down, we were all discussing it internally," said Shashank Agarwal, cofounder of Noveum.ai, who is based in San Francisco. Agarwal warned that Indian startups will lose out on compounding benefits, as using the latest models to write code and build businesses creates momentum that is now disrupted.
Vijay Rayapati, chief executive of Atomicwork, noted that such access restrictions have historically appeared in defense and chip industries. However, AI architecture changes multiple times a year, unlike fighter jets or graphics processing units which update on five- to ten-year cycles. "There is no way to keep up with this acceleration if you don't have access," Rayapati said. Engineers facing these sudden access restrictions may need to rely on structured upskilling, such as an AI Learning Path for Software Developers, to adapt to shifting tool availability and maintain technical relevance.
Why this matters for IT and development professionals
Software engineers can no longer assume uninterrupted access to top-tier foreign AI APIs. Building resilient applications now requires designing systems that can swap foundation models or fall back to localized alternatives without breaking core functionality. Teams must audit their dependencies and prepare for sudden API access revocations based on geopolitical directives.
Your membership also unlocks:
