AI Quote of Note: “Colorado communities are deeply worried about what this unrestrained development means for their water, their air quality, their electricity bills, their farmland and their neighborhoods.”
Who said it: State Sen. Cathy Kipp, D-Fort Collins, as Colorado lawmakers killed her data center regulation bill in May.
Why it matters: For much of the past two years, the public conversation about artificial intelligence has focused on models, chatbots, copyright, bias, jobs, hallucinations, and regulation. Those issues still matter, but AI has also become something truly concrete: a story about data centers, the large facilities that store, process, and move the digital information behind cloud computing, streaming services, business software, and AI tools. As companies race to build more powerful AI systems, the demand for electricity, cooling, land, transmission capacity, and backup power is growing with them.
That's why Kipp’s quote, as reported in The Colorado Sun, deserves attention. It pulls the AI debate out of the screen and into people's neighborhoods.
The competing bills
In Colorado, the 2026 data center fight split into two competing bills. House Bill 1030, sponsored by Rep. Alex Valdez, D-Denver, and Sen. Kyle Mullica, D-Thornton, would have created long-term sales and use tax exemptions to attract large data center projects.
Senate Bill 102, sponsored by Kipp and Rep. Kyle Brown, D-Louisville, started from the other direction: It sought to make large-load data centers cover more of their power and grid costs, meet clean-energy and water-use requirements, report electricity and water consumption, and engage more directly with affected communities.
In the final days of the session, Kipp rewrote her bill to add a limited tax incentive, trying to bridge the gap between economic development and environmental accountability. It still failed. So did the industry-backed incentive bill. As the Denver Post pointed out, the failure of the bills has left Colorado with neither incentives to lure new development nor rules about the centers’ use of power, water and land.
The water question
The concerns are not theoretical. A Colorado legislative memo cited Denver Post reporting that CoreSite’s north Denver data center campus could use up to 805,000 gallons of water a day for cooling, a volume the memo compared to the average daily indoor water use of 16,100 Denver residents.
Since publication of this article, one reader raised a fair point about context: Measuring a data center's water use against household consumption can land harder than a comparison would to other industrial users. To their point, it's worth noting that Colorado has more than 200 golf courses, some drawing hundreds of thousands of gallons a day in peak season, and a single city's summer lawn watering can run well into the millions.
But the comparison cuts more than one way. Much of the water that households and cities use indoors returns to the system as treated wastewater, while the evaporative cooling common at large data centers mostly evaporates and does not come back. Therefore, by the measure that matters most for supply – consumption rather than withdrawal – a water-cooled facility's footprint can be significantly heavier than that suggested by the raw gallon count. And the quantity of gallons themselves vary enormously: A similar-sized campus might use tens of thousands of gallons a day on a closed-loop, air-cooled design, or several million on evaporative cooling.
However, none of this shrinks the concern Kipp and others raised. If anything, it underscores why the kind of disclosure her bill sought really matter, since communities usually have no way to know which design a facility runs.
Call it the black box of the data center: AI taught the public to worry about opaque algorithms, but the buildings that house them can be just as hard to see into. Their real draw on power, water, and land is often invisible to a community until the facility is already running.
Who pays for the buildout?
Ultimately, all of this leaves Colorado in an uncomfortable position. The state wants to compete in the AI economy and attract technology investment, but it also has to answer basic questions about who pays for the infrastructure that AI requires.
If a data center needs major grid upgrades, should those costs fall on the company, the utility, or ratepayers? If a facility uses large amounts of water or relies on backup generators, how should communities weigh those impacts against potential tax revenue and construction jobs? If communities are asked to host the physical backbone of the AI boom, what protections should they have before deals are made?
Those questions are especially important because data centers do not always create the kind of long-term employment their size might suggest. They can produce construction jobs, property tax revenue, and local economic activity. But once built, many data centers operate with relatively small permanent staffs. That doesn't mean they are bad for a community, but it does mean the trade-offs should be understood clearly before public officials offer incentives or approve major projects.
Kipp’s quote is also a useful reminder that AI accountability cannot be limited to model behavior. A chatbot that gives bad advice raises one set of concerns. A data center that strains the grid, affects air quality, increases water demand, or shifts costs to households raises another. Both of them are AI issues now.
From Algorithms to Infrastructure
For Colorado, this may become one of the biggest – if not the biggest – of the state's AI policy fights. Colorado has already been near the front of the AI policy debate, though its approach changed significantly this year when lawmakers replaced the state’s original high-risk AI law with a narrower framework for automated decision-making. Data centers move the conversation into a different arena: energy, climate, land use, utilities, local control, and economic development.
That shift matters. AI may feel weightless when it appears as text on a screen, but the systems behind it are made of buildings, power, water, land, chips, permits, and public trade-offs.
The failures of the two bills didn't end the fight so much as relocate it. Days after the legislative session adjourned, the Denver City Council voted unanimously to impose a one-year moratorium on new data center development, pausing the acceptance and processing of new zoning permits and site-development plans while the city drafts rules on energy use, water consumption, noise, and where such facilities can be built.
Denver had no data-center-specific zoning, energy, or water regulations on the books before the vote. The pause does not stop the CoreSite project already under construction in the north Denver neighborhood, but it blocks the two additional buildings planned for that campus – and Denver is not alone. Aurora has restricted data center water use since 2023, and Jefferson County and Longmont are now writing rules of their own. With the statehouse deadlocked, Colorado's data center policy is increasingly being made one city council at a time.
Kipp has said she plans to bring the data center issue back next session. When she does, Colorado’s debate over AI will likely become more concrete. The question will not only be what AI can do, but what AI demands from the communities asked to power it.
