AI Intelligence Briefing: AI Sovereignty in India, When Competitors Agree, and a Defense Against Voice Scams
India moved to own its AI infrastructure, AI's biggest rivals warned Congress together, and Google launched protection against AI voice scams.
Welcome to your AI Briefing, where I wade through the news chaos so you don’t have to. Every other Friday, we cover three stories: what happened, why it matters, what it means for our lives, and the bottom line.
This week:
India partnered with the UAE to put AI supercomputers on its own territory, bypassing Amazon, Microsoft, and Google to build infrastructure it owns and controls.
The CEOs of OpenAI, Anthropic, Google DeepMind, Microsoft, and Meta set aside their rivalry to co-sign a public letter warning Congress that their own models could help someone build a biological weapon.
Google launched a feature that tells us when a phone call may be AI impersonating a voice we trust.
India and the UAE created a blueprint for countries that want to own their AI, not rent it from American companies
What Happened
For most governments that want AI capability, the path has been to rent computing power from Amazon, Microsoft, or Google. These companies operate the data centers, own the hardware, and govern the infrastructure that most of the world’s AI currently runs on. India already had at least $45 billion in commitments to all three. This month, it signed a different kind of deal. G42, an AI company based in the UAE and backed by Abu Dhabi’s government investment fund, agreed to deploy 64 supercomputers made by US chipmaker Cerebras on Indian soil. (Cerebras builds chips designed for speed in running AI applications, which is different from the Nvidia processors India uses for training models.) The choice reflects India’s focus on deploying AI across healthcare and public services rather than building large foundation models from scratch.
The machines will sit on Indian territory under Indian governance, with G42 handling operations. India is the first country to join what G42 calls the Intelligence Grid, a global network of AI computing facilities it builds and operates for governments. G42 is already in discussions with other governments about the same model. Cameron Kerry, former acting secretary at the US Department of Commerce, reflects that India’s “pragmatic approach to AI sovereignty, using the power of its scale to adapt what’s available from other countries to its own needs.” G42 brings hands-on experience to the arrangement: it built and now operates supercomputer facilities in the US and became Cerebras’s largest customer before expanding into this government-focused model.
Why It Matters
Until this deal, a country that wanted serious AI computing capability had to route that capability through American infrastructure, on American companies’ terms, and subject to their supply chain and regulatory decisions. India’s arrangement with G42 introduces a different option: machines physically on Indian soil, data governed by Indian rules, managed by a partner outside the American technology ecosystem. This is what AI sovereignty looks like in practice. The Intelligence Grid model is designed to replicate this arrangement across other countries, and the fact that G42 is already in discussions with other governments means India’s deal may be the first of many.
What It Means for Us
The AI tools available to us today are products of a specific moment in which a small number of American companies control most of the world’s AI infrastructure. That control shapes what gets built, for whom, and on what terms. If the Intelligence Grid model spreads, the next generation of AI development happens in a more distributed world, with more governments owning the infrastructure that serves their populations and making decisions about how it operates. The tools that emerge from that world will reflect a wider range of priorities than the ones currently shaping what we have access to now.
Bottom Line: India’s deal turns AI sovereignty into a budget question. Building this kind of infrastructure used to require years of national investment in domestic chip production and data center capacity. G42 has packaged it into a commercial service. India just secured what previously required a national-scale construction effort, which lowers the practical barrier to AI independence significantly.
OpenAI, Anthropic, Google, Microsoft, and Meta jointly warned Congress that their own chatbots could help someone make a biological weapon
