Getty Images
The housing market is stuck in a cruel summer. Existing-home sales are hovering near an annual rate of 4 million (up only 0.8% from a year ago), and affordability remains historically low. But while the COVID-19 pandemic-era run-up in prices has taken the blame, another culprit may be contributing to the stagnant market: artificial intelligence.
The AI boom is fueling a frenzy of data centers that are competing with residential homes for capital, power, water, and even sonic space. The result is a housing market stuck in neutral, with the potential for higher bills on the horizon, and neighborhoods kept awake by the hum of chillers designed to keep machines—not people—cool.
The technology that promised progress is instead creating new, invisible pressures on homeowners, just as a new report from MIT suggests that 95% of generative AI pilots at companies are failing. Here’s what homeowners need to know about the hidden side effects of AI on the housing market.
The mortgage rate squeeze: ‘Let them eat compute’
Jason Thomas, head of global research at Carlyle, said that artificial intelligence may be fueling the great mortgage lock-in in a recent research note titled “Let Them Eat Compute.” In it, he argued that the trillions pouring into AI infrastructure are competing directly with housing for capital.
“At 6.58%, rates on U.S. 30-year fixed-rate mortgages reached their lowest level of the year,” Thomas wrote. “But they still feel really high. To speak of a rate-induced ‘affordability crisis’ is no exaggeration. But this crisis resulted from the sharp downward adjustment in rates in 2020–22 rather than their post-pandemic rise.” In other words, pandemic-era lows reset buyer expectations, making today’s “normal” rates feel crushing.
That gap explains why the market is frozen.
“Existing owners with below-market rate mortgages have a strong financial incentive to remain in their home, or rent rather than sell the property,” Thomas noted. Cheap loans saved U.S. households nearly $500 billion in disposable income annually in 2023 and 2024. With so much at stake, owners are clinging to homes financed at 2% to 3%—what he calls households’ “most valuable asset,” even if it’s technically a liability (i.e., debt).
Buyers, meanwhile, face financing costs closer to 6.5%.
“In real estate terms, existing owners apply a 4.1% ‘cap rate’ to the imputed rent of their property while prospective buyers face 6.58% financing costs,” Thomas explained. “The resulting bid-ask spread explains the collapse in home sales volumes.” Translation: Sellers won’t give up cheap mortgages, buyers can’t stretch to cover today’s costs, and transactions grind to a halt.
And the prospects for relief aren’t simple.
“The U.S. finds itself in the midst of a really expensive data center/supercomputer buildout,” Thomas added. “AI-related infrastructure spending consumes hundreds of billions of dollars of capital each quarter. … With individual companies pledging to spend trillions of dollars on AI infrastructure, growth may accelerate from here.”
Add to that government deficits that now consume 1.5 times as much private savings, and the result is clear: “The issue is not whether high rates are ‘crowding out’ interest-sensitive sectors like for-sale housing. Clearly, they are.”
Utilities: When AI meets the grid
The AI buildout is also devouring energy. Training large language models and running hyperscale data centers requires huge amounts of electricity, forcing utilities to expand capacity and, in some cases, launch entirely new power projects.
Google’s recent partnership with Kairos Power and the Tennessee Valley Authority (TVA) illustrates the scale. The companies are planning a 50-megawatt advanced nuclear reactor in Oak Ridge, TN, with a long-term goal of adding 500 MW of new nuclear capacity largely to fuel data center needs.
For TVA, the appeal is stability.
“Nuclear power does protect consumers from fossil fuel price volatility,” Scott Fielder, a TVA spokesperson, explains. “Today, more than 40% of TVA’s generation comes from nuclear—double the national average of around 20%. TVA operates the third-largest nuclear fleet in the country, giving our region a competitive advantage. Nuclear is among the lowest-cost, most reliable, and safest forms of energy available.”
That reliability could spill over to consumers. Under the deal, Kairos Power will deliver enough electricity to power nearly 290,000 homes, with lower-cost and more stable pricing. Even better, ratepayers aren’t footing the bill to build it.
“TVA assumes no development risk and will not pay anything until Hermes 2 begins generating electricity. Even then, TVA will only pay for the electricity it receives—at a fixed, market-based price,” Fielder explains.
It’s a rare silver lining in an otherwise fraught landscape. Across much of the country, grid upgrades and new-generation projects are pushing household utility bills higher. For homeowners, the tension is unavoidable: AI’s energy appetite is enormous, but investments like this one suggest that the scramble to power supercomputers could, at least in part, ease the pressure on families trying to keep the lights on.
Water wars in the AI age
AI is also competing for something even more basic and essential: water.
In Newton County, GA, Meta’s $750 million data center guzzles about 500,000 gallons of water a day, roughly 10% of the county’s entire supply, the New York Times reports.
That’s left some residents such as Jeff and Beverly Morris living with taps that spit out brown sediment. Their dishwasher, washing machine, and even toilets have failed repeatedly since construction began in 2019.
“It feels like we’re fighting an unwinnable battle that we didn’t sign up for,” Beverly said in an interview with the Times, adding that they’ve spent $5,000 on repairs and can’t afford the $25,000 to replace their well.
The county as a whole is projected to face a water deficit by 2030, forcing local officials to weigh rationing and approve a steep 33% rate hike over the next two years, compared with the usual 2% annual climb. For families already stretched thin by housing costs, higher utility bills for something as essential as water are one more burden layered on top of an already difficult market.
A new noisy neighbor: Like ‘somebody in pain, crying’
Noise has become another risk of AI’s expansion. In Chandler, AZ, one resident near one of the country’s largest data center campuses described the constant whine of industrial chillers to The Atlantic in 2019 as “somebody in pain, crying. Crying constantly and moaning in pain.”
The hum, which falls in the low-frequency range that has also been associated with the hair-raising feeling of haunted cellars, has kept families awake, prompting some to consider selling their homes and others to organize a collective.
Their group, Dobson Noise Coalition, has been organizing neighbors since 2018 against the noise from nearby data centers. Members have experienced headaches, irritability, and difficulty sleeping.
In a post on their Facebook page, one member recently said, “Sounds like a jet plane taking off. I normally don’t hear it but tonight is insane.”
There’s little local officials can do to respond. Despite hundreds of complaints, the Chandler Police Department concluded the sound wasn’t “significant enough” to cite under city code.
While noise pollution may be common in a bustling city like New York, it’s less of an accepted fact of life in places like Chandler. That can create a problem for home values. Studies have found that proximity to railroad tracks, highways, and airports can all discount a home’s value by as much as 13%.
Buyers usually weigh those factors when choosing a home. But the boom in data center construction means some homeowners may discover, too late, that their quiet neighborhood now comes with a noisy new neighbor.
What this means for homeowners and buyers
Even if a data center isn’t moving in on your block, you’re likely feeling AI’s impact on the housing market.
Buyers are seeing it in higher mortgage rates, as data center buildouts keep swallowing trillions in capital. Sellers are taking an equal hit, as they get trapped by the mortgage “lock-in” effect. And for homeowners, having a tech giant neighbor is less a blessing than a burden: strained water supplies and sleepless nights filled with industrial hums can chip away at property values and peace of mind.
The irony is hard to miss. AI was supposed to deliver efficiency, productivity, and progress. Instead, for millions of households, it’s delivering higher bills, frozen markets, and hidden costs.