Yesterday’s AI article began with a simple online search: asking for help planning a menu. Millions of people ask something similar every week. But instead of getting a list of sources to help you do the work, today you’ll probably see an “AI Overview.” These are generated in an AI data center.

According to the U.S. Energy Information Administration, a small data center uses 5 to 10 megawatts of electricity, but 100 megawatt “hyperscale data centers” are increasingly common. One megawatt can power 1,000 homes, so a 100 megawatt data center uses the same amount of power as 80,000 homes. Small data centers can be built on 2 to 8 acres, but larger ones can take up 100 or more acres.

Inside data centers are rows of computers and servers that process, store, and transmit data. They have few doors or windows, and are ordinarily built of cast concrete panels. Security is tight with high metal fences, security cameras and motion detectors.

According to a study by The Environmental Research Letters, they estimate a data center uses roughly seven cubic meters of water per megawatt hour of energy. The centers have huge, roof-mounted air-conditioning systems requiring large amounts of water to keep the computers cool.

Data centers of telephone and internet service providers, or insurance companies processing medical claims, use central processing units, or CPUs, but AI data centers require graphics processing units. These relatively new computers have thousands of cores, allowing them to process many inquiries at once. They’re larger and faster than CPUs by an order of magnitude.

In addition to fielding inquiries, AI data centers train computers to improve the accuracy of text and realism of graphic images. This is where computers become more autonomous.

Reacting to negative publicity surrounding the amount of electricity used by data centers, Microsoft, Amazon and Google have announced plans to invest in nuclear energy to power them. Until then, today’s utilities, most powered by coal or natural gas, must turn out more and more electricity for AI every month.

According to a report from the Lawrence Berkeley Laboratory, in 2023, AI consumed an estimated 4.4% of all electricity in the U.S. Some projections show that could double or triple by 2028.

According to Governing Daily, Virginia and Texas have the most data centers, with Virginia well over 600.

Iowa is proving attractive for new data centers due to relatively low electrical rates and abundant wind energy.

Today, two large data centers are under construction in Alliant Energy’s Big Cedar Industrial Center near Cedar Rapids: one by Google, the other by QTS Data.

In February 2026, the Linn County supervisors recently took steps to regulate data centers’ use of electricity and water, according the Linn County supervisors minutes. At that time, Google began negotiating with the city of Palo to build a new six-building data campus, ostensibly to avoid the regulations. The new center requires 12 million gallons of water daily, and it’s unclear what contaminants it might contain when returned to the Cedar River.

Elsewhere, Meta expanded its 5 million square foot data center in Altoona, which it calls its largest in the world. In West Des Moines, Microsoft is building Project Ruthenium on 123 acres, its sixth facility in the area. The company says it will be 100% powered by renewal energy, and create 25 new jobs.

We asked members of the City Council of Webster City how they’d feel if a data center expressed interest in locating here.

Mayor John Hawkins, said he would consider any economic development proposal strictly on its merits, and would not categorically rule out a data center. Councilman John Marvel said, “It’s a hypothetical question, but I’d certainly look at it from every angle before making a decision.” Councilwoman Abbey Hansen, said she wouldn’t make up her mind “on the spot,” saying “it’s something we’d need to look into.”