Behind the rapid integration of AI into society, enormous data centers are consuming electricity and water on the scale of cities. They bring with them major issues such as pollution, draining of public resources, and surveillance.
In this conversation with the World Council for Health, privacy advocate and Above Phone founder Hakeem Anwar warns that the scale of this buildout is being dramatically underestimated.
Hakeem Anwar has produced a map of data centers and their estimated impact on the environment in the US. This is an important resource for us to begin taking action to prevent their expansion.
AI Is Becoming an Infrastructure Problem
Traditional internet services have long depended on data centers, but Anwar emphasizes that AI is driving infrastructure to an entirely different scale. Hyperscale facilities can require 100 megawatts or more, while emerging gigascale facilities can exceed one gigawatt—roughly equivalent to the electricity consumption of 800,000 to one million homes.
Enormous Demands for Water
Data centers also require enormous amounts of water for cooling. Anwar’s map estimates that one facility could consume approximately 2.5 million gallons of water per day, while a 50-mile cluster of data centers around Phoenix was estimated at roughly 17 million gallons per day.
He stresses that these are estimates based on published research and assumptions about facility operation. Yet this uncertainty highlights part of the problem: Anwar reveals that data-center sustainability reports he examined were themselves estimating rather than directly measuring water consumption.
Pollution and the Cost of AI
The issue extends further. Anwar highlights that gas turbines and diesel backup generators are being increasingly used as rapidly expanding data-center infrastructure places additional demands on electrical grids.
Anwar also raises concerns about the refrigerants and coolants needed to control the tremendous heat generated by AI hardware, particularly how these chemicals are handled and potentially returned to surrounding water systems.
Why Are Data Centers Being Built So Quickly?
Anwar points to substantial federal, state, and local tax incentives as one factor accelerating construction. While enormous private investment pours into new facilities, communities may ultimately bear some of the infrastructure costs associated with supplying them with electricity and water. The extraordinary speed of the buildout makes public scrutiny essential and urgent.
The AI Data Center Map
To expose the scale of this infrastructure, Anwar developed the AI Data Center Map, allowing people to locate nearby facilities and examine estimated electricity and water consumption. The project is collaborative: users can submit corrections as projects shift, expand, or are cancelled.
The Infrastructure of Pervasive Surveillance
Anwar also warns that enormous AI computing capacity could provide the infrastructure for a new generation of intrusive surveillance. Cameras, license-plate readers, smartphones, doorbells, and laptops already generate immense quantities of information. Connecting these streams through AI, and potentially tying them to digital identity, could enable increasingly comprehensive monitoring of people’s activities.
Conclusion
AI requires enormous buildings, chips, transmission infrastructure, cooling systems, electricity, and water. As hundreds more facilities are planned, far greater scrutiny is needed around their resource consumption, environmental consequences, public subsidies, and ultimate purpose.