The US government’s recent announcement strictly restricting access to Anthropic’s advanced AI models Mythos and Fable only to approved American entities and US citizens sent not just several organisations but even countries scrambling. Interestingly, even the foreign employees of Anthropic who may have contributed to building these models were prohibited from access. So, is America being unfair? Should technology remain open?

Important questions, but the real issue for us is why India finds itself vulnerable every time technology becomes strategically vital.

India’s digital journey has been characterised by strong participation while ownership lies elsewhere. In the 1990s, we became enthusiastic users of personal computers while Microsoft and Intel controlled the operating system and the CPU. In the 2000s, we embraced the internet while Google owned search and Cisco controlled the plumbing. In the 2010s, we became one of the world’s largest social media markets, yet Facebook owned the platforms. We adopted smartphones at an astonishing scale but all the value continues to accrue to Apple, Samsung and others. Cloud computing transformed businesses across India while the infrastructure was built and owned by Amazon, Microsoft and Google.

Now once again, in the AI era, India risks becoming a consumer rather than a creator. Obviously, talent is not the challenge because every major global technology company has Indian engineers in leadership positions and Indian researchers contribute to the world’s most advanced AI systems. So, why does ownership still elude us?

Part of the answer lies in the success of our IT services industry. Infosys, TCS, Wipro and others created an extraordinary industry, generated employment and global credibility, and India should rightfully be proud of this achievement. But this success has a downside. We became experts at human arbitrage, integration and customisation, and built businesses around helping others use technology without building our own technologies. The AI boom is exposing this weak spot.

Hundreds of Indian AI startups are doing excellent work but most depend on models developed by OpenAI or Anthropic, which is fine so long as access remains open. But what happens when access becomes geopolitical and national priorities outweigh commercial interests?

There’s a weak argument that India need not build foundational models and instead focus on applications using AI platforms. History suggests that companies that own operating systems, control search engines or build cloud infrastructure capture more value than those using them. AI will likely follow the same pattern, although, to be fair, India still doesn’t have the ecosystem maturity to build an OpenAI.

Ironically, India had an opportunity to be close to the centre of AI action. Infosys was among the early investors in OpenAI and could have been in pole position today, but unfortunately exited way too early — yet another ship that sailed before we realised where it was headed. The Mythos-Fable restrictions are the latest reminder that in technology the greatest risk is in becoming dependent. In the next column, we will discuss how India can try and overcome this dependency.

(The writer is a serial entrepreneur and best-selling author of the book ‘Failing to Succeed’; posts on X @vaitheek)

Published on June 22, 2026