Badgered State: AI data centers and what it means for the state

This is Part 2 of a 3-part series from State Sen. Chris Larsen

By Chris Larson

(In Part 1) we mostly discussed the negatives of AI data centers – and we’re not done yet. But, in the interest of fairness, we should also point out who benefits. This is especially useful in understanding why there is such heavy momentum for these projects despite strong and growing opposition from the public.

The Winners

Construction trades

One clear beneficiary of AI data center projects in the short term is the construction trades. A single data center project can include up to 1,000 construction jobs during the initial build. Despite Wisconsin having repealed its prevailing wage law, many of the firms building AI data centers use union labor to do it. These are good-paying jobs for a variety of skilled trades professions. However, each individual project negotiates its own labor agreements, so depending on the developer, the percentage of union and Wisconsin-based jobs could vary widely.

Big tech companies

The biggest winners when data centers are built are the shareholders and executives of the big tech companies. Whether it’s Oracle, Microsoft, Amazon, Google, Meta, or any number of other tech giants, AI is central to their business, and if they are able to expand their footprint quick enough to edge out would-be competitors, their stock price goes up and wealthy shareholders (including their top executives) get very, very rich.

Private utilities

After the tech giants, for-profit electric utilities like WE Energies stand to gain the most from AI data centers, for the simple fact that every large data center requires obscene amounts of electricity to run. How much power? Well, a September report from Clean Wisconsin found that just 2 large data centers - in this case the Microsoft project in Mount Pleasant and the Oracle/OpenAI project in Port Washington - will consume as much electricity as every household in Wisconsin combined and then some. They are predicted to require 3.9 gigawatts of electricity each year, which could power 4.3 million homes. There are currently only 2.8 million households in our state.

More electricity usage means new, expensive power plants. Because the Public Service Commission (PSC) granted WE Energies a near-guaranteed 9.8% rate of return, the more money that is spent building new power plants, the richer their shareholders get (and the higher your bill will go). 

Local governments

Finally, local governments stand to benefit from AI data centers due to increased property tax revenues. Take the Microsoft data center in Mount Pleasant. Assuming their $3.3 billion investment received a tax assessment of a similar value from the Village, at the 2024 mill rate of $16.09 per $1,000 of assessed value, Microsoft would have to pay $53.1 million per year in property taxes. What they actually end up paying remains to be seen, and could be significantly lower if additional incentives are provided by the Village, but considering that Microsoft’s data center could increase the property tax base in Mount Pleasant by almost 50%, it’s understandable why they and other local governments would want a data center in their community. Homeowners would also likely see a reduction in their property taxes in the short term as the tax levy gets spread out among a much larger tax base.

It’s not all good news on the property tax side, however. Having one property taxpayer represent such a large percentage of the total amount collected automatically gives that taxpayer an outsized voice in the decisions made at the local level. We saw this in Milwaukee during the last school referendum when Northwestern Mutual’ former CEO spent big bucks opposing local efforts to make up the funding gap left by the state. Northwestern Mutual is the largest property taxpayer in Milwaukee County, but their equalized property value of about $724 million is peanuts compared to a multibillion dollar data center. Do we really want to give multinational tech conglomerates the ability to dictate which schools get closed, what teachers are paid, or how many plow trucks are on the road during the next winter storm?

The losers

Now that we know who the big winners are, it’s time to take a look at who loses. In short, it’s just about everyone not mentioned in the previous section - which is the vast majority of the population. This manifests in three major areas: electric rates, our environment, and job elimination.

Electric rates

On average, Wisconsin had the second-highest residential electric rates of the 7 Midwestern states (IA, IL, IN, MN, MI, OH, WI) according to recently-released 2024 data from the U.S. Government’s Energy Information Administration (EIA) - at $.1272 per kWh. Customers of large utilities like WE Energies (with over 1 million customers) tend to pay significantly higher rates than those who have municipal power utilities or co-ops. In 2025, for instance, WE Energies customers pay $.1833 per kWh, while customers of municipal or co-op utilities can pay less than $.10 per kWh. Residential customers also tend to pay much higher rates than large business customers, so any shock to electricity prices will be felt almost immediately by families.

With all this in mind, the prospect of even one or two hyper-scale data centers is a frightening prospect indeed. We already mentioned that just 2 of these proposed projects would require more power than every household in the state combined. If we see 8 or 10 of these mega projects come online in the next decade, we’ll need to build a bunch of new power plants, which - as I mentioned previously - will be great news for utility investors and construction jobs but terrible for the rest of us.

Our environment:

Reliance on fossil fuels

Of course, it’s not just electric rates that must be considered when a massive boost in electricity consumption occurs, it’s how that power is generated. Sadly, Wisconsin is lagging behind the rest of the nation when it comes to renewable energy. WE Energies - the state’s largest electric utility (and the likely provider of power for multiple large data center projects) - obtained just 5.3% of its 29.2 million MWh of electricity from renewable sources in 2024, which is actually a DECREASE from 6.2% in 2022. The rest of Wisconsin is doing somewhat better. According to EIA data from August of 2025, our state gets 12.7% of its electricity from renewable sources. Notably, this is still well behind the national average of 20.1%, and worlds behind leaders like South Dakota, which got 69.8% of its power from renewables in August.

Why does all of this matter? Well, besides renewable energy sources being cheaper on average than fossil fuels, they carry much less risk of harmful air quality effects than coal or methane gas, which together make up almost two-thirds of WE Energies’ electric capacity.

How harmful can coal-burning power plants be for local air quality? Well, according to a report from Physicians for Social Responsibility, “Reducing air pollution from fossil fuel facilities can deliver immediate and measurable health benefits. For instance, when a coal coking plant in Pittsburgh, Pennsylvania shut down, nearby communities saw a rapid decline in emergency department visits - over 20% fewer for respiratory symptoms and more than 40% fewer for pediatric asthma.”

As someone who has suffered from asthma since I was a little kid, I know firsthand how harmful fossil fuel infrastructure can be to people’s health. 

Now, there’s nothing saying that AI data centers must be powered by fossil fuels. However, unless specific mandates are put in place by the state, there’s no reason to believe Wisconsin utility companies - particularly WE Energies - will suddenly rapidly shift their power generation ratios to be more favorable for renewables. WE Energies has a history of loudly making big pledges about transitioning to renewables but then quietly giving up and sticking with fossil fuels instead. The only way to ensure WE Energies - or any investor-owned utility - makes good on their green energy promises is through legislation requiring them to do so. .

Our water

We already established that most data centers use a ton of water, and perhaps unsurprisingly, the companies responsible often understate water needs when seeking approval for them. This water is needed for the cooling process (which can vary heavily by the type of cooling system used), as well as the power generation itself. As such, a data center’s water needs can fluctuate dramatically based on the type of power. 

For example, according to environmental watchdog group Clean Wisconsin, the 3.5 Gigawatt data center proposed for Port Washington could require 54 million gallons of water per day if powered by a methane gas plant - enough to satisfy typical daily water needs of 970,000 people. That’s more than the entire population of Milwaukee County, by far our state’s most populous county. If this data center were to be powered by nuclear energy, however, the 54 million-gallon estimate above could increase more than sixty-fold - to 3.7 billion gallons per day.

There’s also the risk of water being contaminated in the cooling process. These contaminants can be chemical - including biocides, corrosion inhibitors, and refrigerants, biological - including bacteria and algae, and heavy metal - including copper, zinc, and even lead. With all the issues our communities here in Wisconsin have had with “forever chemical” contamination (also known as PFAs), why would we want to risk contaminating our water supply even more?

Job elimination

Do you ever wonder why your phone starts to run worse when a new software update comes out? Or why your old charger no longer has enough juice to power your new tablet? One term to describe this phenomenon is “planned obsolescence.” Well, the endgame for AI as a technology - which is only fueled by the explosion of data center buildouts across the country - is a different kind of planned obsolescence. At its core, it’s us - or rather, our jobs.

It’s definitely annoying when you feel compelled to replace an otherwise working TV, game console, or Bluetooth headset because the latest features won’t work on them, but none of us are quite prepared for how profoundly AI-related job loss will affect our lives and livelihoods over the next few decades if big tech companies and their corporate clients have their way.

In October, the New York Times reported that Amazon - which employs about 1.1 million people in the United States - announced it plans to automate 75% of its operations and replace over half a million jobs with robots.

Also in October, Democrats on the U.S. Senate HELP Committee released a report which found that AI could eliminate up to 97 million jobs over the next decade, including 89% of fast food and counter workers, 64% of accountants, and 47% of truck drivers. Workforce consulting firm McKinsey found in 2023 that AI could automate tasks that currently take up 60-70% of employees’ time.

Even if AI technology was not at risk of killing millions of jobs, it’s relevant to note that even for the largest hyperscale data centers, the amount of full-time permanent jobs created is quite small. Even for a data center that might employ 1,000 construction workers in the short term, there might only be about 100 full-time permanent jobs created, and there’s no guarantee at this time that any of them would go to existing Wisconsin residents. Meta, Google, Oracle, et al. could just as easily fly in a few workers from out of state, which wouldn’t address any of our communities’ employment needs - particularly in traditional farming communities that are struggling to transition to the modern economy as family farms close and consolidation reigns.

Knowing all of this, and knowing that the current President and Congress have no plan to create a soft landing for folks whose jobs are in danger of elimination, why would we want Wisconsin to facilitate job-killing AI with open arms?

Part 3 tackles the unknowns of AI