Amazon AI data centers in India face backlash over water and power strain

This story captures the physical side of the AI boom that is often ignored in model-centric debates. India is positioning itself as a global AI infrastructure hub, with Amazon, Microsoft and Google committing tens of billions of dollars to data centers. But you cannot scale “the world’s data” without power, water and land, and those are exactly the resources local residents fear losing. The protests around Amazon’s Thane facility echo tensions already visible in Virginia, Dublin and Amsterdam, suggesting that social license for hyperscale AI infrastructure is far from guaranteed.

From an AGI perspective, this is about bottlenecks moving from GPU counts to public patience. Compute availability is the main hard constraint on training the largest models, and countries like India look attractive because of cost and political appetite for investment. If public backlash forces projects to relocate, add mitigation costs or face lengthy approvals, the effective price of compute rises. That does not stop AGI research, but it can slow the most aggressive scaling plans and push companies toward more efficient architectures.

At the same time, governments see AI data centers as strategic industrial assets. The Modi administration’s bet that owning the “physical backbone” of AI can rival leadership in frontier models is a credible alternative path to influence. How India balances that ambition with environmental limits and community consent will shape where global labs and cloud providers site their next generation of AI factories.