Environmental destruction will accelerate
AI’s physical footprint is growing with its capabilities, consuming enormous amounts of energy, water, and land as companies race to build larger data centers and use more power.
Here's a bipartisan issue for you: almost nobody wants to live next to a data center. In a recent survey, 75% of Americans would oppose a new data center near where they live, with more than six in 10 strongly opposed.1 We predict that these numbers would rise if this situation was taken from a hypothetical to an actuality. But to build a superintelligence, these labs will need many more of them all the same.
Electricity use
Electricity demand is surging, and AI is the primary driver. The IEA’s 2025 Energy and AI report projected that global data center electricity demand will more than double by 2030 to about 945 TWh, with demand from AI-optimized data centers projected to more than quadruple.2
Even the labs working on these technologies acknowledge the cost to the environment. Google recently reported that its greenhouse gas emissions had risen 48% since 2019, attributing the increase to data center energy use and supply chain emissions.3
Water use
Unfortunately, the problem extends far beyond electricity usage. A Harvard analysis of 472 large U.S. data centers estimated about 300 billion liters of operational water use per year, roughly three-quarters of it from electricity generation rather than on-site cooling.4
E-waste
E-Waste also exists as an emerging problem, due to the need to upgrade hardware as models become more powerful. Many components only last 2-5 years, and the rate of replacement has accelerated. A study in Nature Computational Science estimated that without regulatory changes, generative AI alone could produce 1.2–5.0 million tonnes of e-waste by 2030.5
Community pollution
The cost of this environmental destruction falls on the very communities that are opposing it. Consider the expansion of xAI in Memphis, where the company installed at least 18 gas combustion turbines, each of which can emit 130 tons of nitrogen oxide per year.6 This took a bad situation and made it worse, as the county has received an "F" air quality grade from the American Lung Association.
Considering these points in totality, a disturbing picture emerges: in their race to build a superintelligence, AI labs are also accelerating the destruction of our shared environment, while the executives isolate themselves from the consequences.