AI, in all its hyped glory, has induced changes into almost every part of the information technology (IT) sector. We’ve seen NVIDIA’s value skyrocket over its market-leading GPUs for AI. We’ve seen Intel take a tumble over its failure to capitalize on the growth in AI chips. And we’ve seen every IT sector vendor add the term ‘AI’ to seemingly every solution.

“AI mania” has spurred a boom in data center construction that has environmental, governmental, and energy experts concerned. AI has caused the need for data centers to skyrocket, and those data centers are by necessity far more power hungry than those of the past. These demands occur while electric automobiles are sapping more electrical energy from already over-burdened power grids and concerns about climate change continue to grow.

The problem is far bigger than just power consumption. Data centers require tremendous amounts of water to keep them cool. As CPUs and GPUs (plus all the other supporting silicon) have grown more complex, the amount of power they use has risen dramatically. That also means that the amount of heat that must be moved away from the data center has increased accordingly.

Government regulation has already descended, with cities restricting data center construction and governments eyeing even more regulation to address environmental concerns. Companies and investors that have strong commitments to green principles are less inclined to invest in generative AI (GenAI) or data center construction. Further, concerns about supplies and environmental effects of rare mineral resources also weigh on the minds of governments, investors, and customers.

There isn’t much that can be done about the status of power generation and distribution – renewable power solutions and grid expansion are projects that take years to implement. New and innovative ways to dissipate heat in data centers and repurpose that heat to other uses are being developed. In theory, GenAI itself should grow more efficient over time. None of that will arrive soon enough or be efficient enough for the projected proliferation of data centers that are needed to support the growth projections for GenAI.

At the risk of screaming into the void, data center growth must be slowed until solutions for energy generation, distribution, and other environmental concerns catch up to the need. The headlong FOMO-driven rush to build as much GenAI capacity as possible is both self-destructive and unsustainable, for the environment and for the GenAI industry. Regulation and outrage will pop the AI bubble unless there is some restraint.