What Does Your ChatGPT Prompt Have to Do With Your Electric Bill?

by Claire L. Brady

If you have been following the conversation about artificial intelligence lately, you have probably heard some version of this: But what about the data centers?

Maybe you have seen headlines about enormous electricity use. Or water consumption. Or communities pushing back against new facilities. Perhaps you have heard claims that asking ChatGPT a question requires some alarming amount of water or electricity.

It can be hard to separate legitimate concern from internet hyperbole.

So, what should higher education leaders actually understand about data centers?

Physical Infrastructure

First, AI does not happen somewhere in the magical cloud. The models and tools we use depend on physical infrastructure: servers, buildings, cooling systems, electrical equipment, transmission lines, and the energy required to run all of it. As AI use grows, so does demand for that infrastructure. Even Bill Gates recently acknowledged that communities are raising concerns about the energy and water required by data centers as part of the larger societal conversation about AI.

Electricity Use

And the electricity question is real. After years of relatively flat electricity demand in the United States, consumption is growing again, with data centers among the major drivers. The U.S. Energy Information Administration expects increasing demand from large computing facilities to continue contributing significantly to electricity growth, while federal regulators are already working through how these very large new customers connect to the grid without compromising reliability or shifting unfair costs to residential customers.

That last part matters. The concern isn't simply, "Data centers use lots of electricity." Lots of things use lots of electricity. The harder questions are whether local grids can add enormous new loads quickly enough, what new generation and transmission infrastructure will be required, and who pays for it. FERC is now explicitly addressing protections intended to keep residential customers from being stuck with infrastructure costs created by large new energy users.

Water Use

Water is more complicated than many headlines suggest. Data centers can use water for cooling, but their consumption varies dramatically depending on cooling technology, climate, server efficiency, local electricity generation, and other factors. Lawrence Berkeley National Laboratory researchers found differences of more than 10,000-fold in water use across data-center workloads. In other words, there is no honest single answer to "How much water does AI use?"

Fossil-fuel Generation

There is also a broader environmental question. If rapidly growing electricity demand is met by additional fossil-fuel generation, AI growth can contribute to increased emissions. EIA modeling suggests faster-than-expected data-center demand could increase fossil generation in some regions, particularly when new demand arrives faster than cleaner generation and grid capacity can be added.

None of this means data centers are inherently bad. Newer cooling technologies can substantially reduce water consumption, companies can invest in new generation and grid infrastructure, and data centers can bring jobs, tax revenue, and economic development to communities. The real story is more complicated than either "AI is an environmental disaster" or "there is nothing to worry about."

And that is precisely why higher ed leaders should understand it.

What should Higher Ed Leaders Know?

As colleges and universities develop AI strategies, sustainability cannot live in a completely separate conversation. Procurement decisions can include questions about energy and water. Sustainability leaders should have a seat at AI planning tables. Institutions can ask vendors about the infrastructure behind their products rather than treating "the cloud" as an invisible place where computation simply happens.

Higher ed leaders are also increasingly the ones people turn to for a levelheaded read on AI, precisely because you're neither the vendor selling it nor the activist opposing it. You have the credibility to hold two things at once: AI is genuinely useful for your students and your institution, and the infrastructure behind it has real costs that deserve honest accounting. You don't need to become an energy policy expert. You need to be able to say, clearly and without defensiveness, "here's what's true, here's what we're watching, and here's how we're thinking about it." That's leadership people can actually trust and it's a far stronger position than either cheerleading or panic.

The cloud, it turns out, is very much on the ground.

This image was created using ChatGPT