Housing Turnover Hits 30-Year Low as Rates Near 7%, So Americans Are Asking AI to Find Them a Home

The housing market isn't just slow. It's frozen solid, and the guy who runs finances at Home Depot has the numbers to prove it.

The average 30-year fixed mortgage rate hit 6.95% for the week ending September 17, up from 6.76% the week before and 6.26% a year ago, according to Freddie Mac. That's crawling back toward the psychologically brutal 7% line that's kept buyers on the sidelines for years.

Home Depot CFO Richard McPhail put a number on just how bad it's gotten. "We continue to operate in what I call 'frozen housing market conditions,'" he told CNBC. On the company's earnings call, McPhail said housing turnover, the share of homes that actually change hands, "has never been lower as a percentage of the housing stock." He noted that historically, whenever turnover dipped near 3% of the housing stock, it bounced back quickly. Not this time. "We've seen housing turnover at these low levels for four years now," McPhail said.

A Redfin analysis backs him up: only 2.8% of U.S. homes changed hands in the first nine months of 2025, the lowest rate in at least 30 years. Homeowners who locked in 3% mortgages during the pandemic have zero incentive to sell into a 7% market. Buyers, meanwhile, are stuck with both high prices and high borrowing costs. Then-Fed Chair Jerome Powell flagged this exact lock-in problem back in September 2024, calling the market "in part frozen." Two years later, according to McPhail, it still hasn't thawed.

Enter the chatbot

Stuck in a market this brutal, a lot of Americans are looking for any edge they can get, and increasingly that means AI.

A LendingTree survey of more than 2,000 adults, conducted in July and released this week, found 72% of respondents would use AI for at least one task in buying or selling a home. More strikingly, 37% said they'd let AI handle the purchase with "minimal human involvement."

The most popular uses aren't glamorous. They're budgeting problems. Homes.com reported that 35% of respondents want AI to search for homes within budget, 28% want it to estimate a home's value, and 24% want help finding down payment assistance. Explaining mortgage options (18%) and comparing offers (16%) also ranked well ahead of tasks like scheduling showings (8%) or negotiating offers (7%), according to Scotsman Guide's breakdown of the same survey.

Younger buyers are leading the charge. Homes.com's separate New Construction Market Landscape Study, done with Good Run Research and CMI Research, found 56% of millennials and 52% of Gen Z buyers used AI tools to compare homes, compared to just 17% of baby boomers. Bank of America's 2026 Homebuyer Insights Report found a similar pattern, with 57% of AI-using buyers leaning on it mainly to estimate affordability and closing costs.

But LendingTree's own chief consumer finance analyst, Matt Schulz, is telling people not to hand over the keys. "AI should be a starting point, not the final word," Schulz said. "A mortgage or real estate professional can help explain trade-offs, local market realities, loan rules and risks that a chatbot or online tool may miss." He added that when a decision "involves an offer, sensitive data or a life-changing amount of money, people still want human judgment and a chance to verify what they're being told."

Real estate agents themselves aren't hiding from the tech. A National Association of Realtors survey found 92% are using AI or plan to, with 71% citing time savings as the draw. But 63% of those same agents flagged accuracy as their top concern, ahead of compliance issues (49%) and misread market data (47%).

That concern has merit. Forbes technology writer Scott Francis pointed out that AI models are only as good as the data feeding them, and bad inputs get baked into a model's behavior at scale rather than showing up as an obvious glitch. In a market where an AI tool is estimating what your house is worth or whether you can afford the mortgage, a bad data feed isn't an abstract IT problem. It's real money.

The other side of the AI boom

While buyers wait on the market to move, some homeowners are already cashing in on AI in a completely different way. Loudoun County, Virginia, the world's densest concentration of data centers, is using the AI-driven building boom to cut property taxes. Jared Walczak of the Tax Foundation told Fox News Digital that data centers generate 45% of the county's local tax revenue while occupying just 3% of its land, saving the average homeowner about $5,800 a year. That's a real number for a real place, even as data center growth nationally has become a flashpoint in the 2026 midterms over electricity costs and water use.

Meanwhile, the loudest voices in AI itself are fighting over how scared everyone should be. Nvidia CEO Jensen Huang told CBS News there is "0% chance" AI ends humanity by 2030, calling doomsday predictions from critics like former Anthropic researcher Jacob Coxon "irresponsible." Huang argued existing liability law, not new AI-specific rules, already covers most of the risk, a position that lines up with the Trump administration's resistance to new AI guardrails.

None of that debate is going to unfreeze the housing market. The rate is still 6.95%. Turnover is still stuck at a 30-year low. And whether an algorithm can actually help a first-time buyer close on a house faster than a stubborn seller and a nervous lender remains an open question nobody in these surveys has actually answered yet.