Can AI save mortgage rates?

For the better part of three years, anyone watching the Federal Reserve has been in the business of waiting. Waiting for inflation to return to 2%. Waiting for the next rate cut. Waiting for the conditions that would finally unlock the housing market, refinance boom, or broader easing that borrowers and lenders have been anticipating since the Fed began its tightening cycle in 2022.

Now a different kind of argument is gaining traction in financial circles - one that bypasses the Fed entirely. Its central claim: that artificial intelligence, deployed at scale across the economy, could prove more powerful in bringing down prices than anything a central bank can do with interest rates. "It's almost like AI is your monetary policy," said Mike Hunstad, who heads Northern Trust's $1.4 trillion asset management division, in a recent interview with the Financial Times. "And it's going to be more effective than anything the Fed or really any central bank around the world can do."

The statement is deliberately provocative. But it lands in the middle of a genuine and consequential debate - one that touches the Fed's next leadership, the path of mortgage rates, and the economics of lending itself.

The supply shock argument

Hunstad's argument rests on a concept economists call a positive supply shock - an event that dramatically increases the productive capacity of the economy, allowing more goods and services to be delivered at lower cost. When that happens, prices fall not because demand is being suppressed by high interest rates, but because the economy is simply producing more efficiently.

Read next: Oil crisis could push back rate cuts, warns Fed's Goolsbee

The historical template he and others invoke is the 1990s, when the widespread adoption of computing and internet technologies transformed business operations across virtually every sector. Productivity surged. Inflation remained subdued even as unemployment fell to multi-decade lows. Alan Greenspan, then chairing the Federal Reserve, took the celebrated - and at the time controversial - decision to hold rates steady rather than tighten, betting that the technology wave was structurally changing what the economy could deliver without overheating.

"If even a portion of those [AI efficiency gains] actually materialise on an economy-wide basis," Hunstad told the FT, "it could be one of the biggest positive supply shocks we've ever seen." The implication for inflation is direct: if AI delivers the productivity gains that its proponents forecast, companies would be able to produce more at lower unit costs, easing the price pressures that have kept the Fed on hold and mortgage rates elevated.

Hunstad's position finds powerful reinforcement in Washington. Donald Trump's nominee to succeed Jerome Powell as Fed chair, Kevin Warsh, has described the AI boom as "the most productivity-enhancing wave of our lifetimes - past, present and future." Warsh argues explicitly that, as in the Greenspan era, the Fed should incorporate the expected disinflationary benefits of AI into its policy framework now rather than waiting for the data to arrive - allowing rates to fall further and faster than current projections suggest. "Everything technology touches gets cheaper," Warsh said in a 2025 interview. "If a central banker waits until the data shows an increase in productivity, my view is you're backward-looking, you're going to be late."

Why the Fed is holding back

The Fed under Powell has resisted the argument - and not for want of interest in AI. Several Fed governors have given detailed speeches on the technology's

potential. The issue is the timing mismatch at the heart of the AI-and-inflation debate: the costs come first, and the benefits come later.

Fed vice-chair Philip Jefferson has articulated the concern in several appearances. Yes, AI productivity gains could eventually lower unit costs and reduce price pressures, Jefferson has said. But in the near term, the investment surge that AI requires - in data centers, energy infrastructure, semiconductors and the workers who build and operate them - is itself inflationary. It competes for land, energy and capital with other productive uses. "A more immediate increase in demand associated with AI-related activity could raise inflation temporarily," Jefferson said at a Brookings Institution event in February, "absent offsetting monetary policy actions." In other words, the economy pays the inflationary bill before it collects the productivity dividend.

Read next: Consumer confidence in AI is falling. Why that matters to mortgage brokers

That view is broadly shared among professional economists. A snap poll by the University of Chicago's Clark Center and the Financial Times found that nearly 60% of top economists say AI's impact on inflation and borrowing costs over the next two years will be close to zero. Most respondents expected AI to reduce the Fed's preferred inflation measure by less than 0.2 percentage points over the next 24 months - a rounding error, not a regime change. "I don't think it's a disinflationary shock," said Jonathan Wright, a Johns Hopkins economist and former Fed staffer.

60%

of top economists say AI's near-term inflation impact will be close to zero

U. of Chicago / FT poll, Feb 2026

1.5%

Projected AI boost to GDP by 2035, rising to 3% by 2055

Penn Wharton Budget Model, 2025

74%

of AI's economic value captured by just 20% of organisations

PwC AI Performance Study, April 2026

The 1990s parallel - and where it breaks down

The Greenspan comparison is seductive, and worth examining carefully - because its lessons are more nuanced than either side tends to acknowledge.

In the 1990s, Greenspan did not cut rates based on optimism. He held off on raising them - a meaningfully different and more modest policy move. Even then, as Oxford Economics chief US economist Michael Pearce has noted, "it wasn't an argument for cutting rates into accommodative territory." The federal funds rate actually rose from under 4% in late 1993 to 6.5% by mid-2000, even as productivity climbed. Inflation didn't fall because Greenspan loosened policy. It stayed low partly because productivity was genuinely rising, and partly because the Fed's credibility kept long-run inflation expectations anchored.

Read next: Why AI is redefining mortgage underwriting efficiency and the path to scale

Moreover, the productivity gains from computing technology took nearly a decade to show up in aggregate statistics. Robert Solow's famous 1987 quip - "you can see the computer age everywhere but in the productivity statistics" - described a real phenomenon: the economy-wide benefits of transformative technology diffuse slowly, unevenly and unpredictably. The Penn Wharton Budget Model projects that AI will increase productivity and GDP by 1.5% by 2035 and nearly 3% by 2055. Those are material numbers over decades. They are not the stuff of monetary policy decisions over the next eighteen months.

A further complication is the current economic context. The Fed left rates unchanged in its March meeting as policymakers confronted the inflationary pressures of the war in Iran, which has driven energy prices sharply higher. The 30-year fixed mortgage rate stood at 6.30% as of April 16, according to Freddie Mac - still far above the sub-3% era that millions of existing homeowners are unwilling to trade away. Austan Goolsbee, president of the Chicago Fed, put the near-term case for caution plainly in February: "You want to be extremely careful. You can overheat the economy easily. Let's be a little bit careful, circumspect."

The productivity paradox: evidence on the ground

Part of what makes this debate so difficult to resolve is that the evidence at the firm level and the evidence at the macroeconomic level are pointing in different directions.

At the firm level, the case for AI-driven efficiency is real and documented. Controlled studies have found substantial productivity gains for legal analysts, software engineers, management consultants and call-center workers using AI tools. A February 2026 report from the International Center for Law and Economics reviewed the empirical literature and concluded that genuine productivity "uplift" is measurable in specific occupational categories. Crucially, a PwC study published in April 2026 found that 74% of AI's economic value is being captured by just 20% of organisations - those

that are using AI as a catalyst for business reinvention and new revenue, rather than simply layering it onto existing processes.

Read next: HSBC's AI gamble puts 20,000 banking jobs on the line

At the macroeconomic level, the picture is murkier. A study by the National Bureau of Economic Research, surveying thousands of chief executives and chief financial officers in the US, UK, Germany and Australia, found that the vast majority see little impact from AI on their operations so far. Productivity climbed at a healthy 5.2% annualised rate in the third quarter of 2025, before pulling back to 2.8% in the fourth. Unit labour costs rose 2.8% in that final quarter - not the pattern one would expect if AI-driven efficiency were already flowing through to restrained business costs and lower prices. The San Francisco Fed captured the tension in a February 2026 Economic Letter: "There's a lot of hope that we've entered a period of higher productivity growth, maybe spurred by AI. My work suggests that remains unlikely" - over the near term, at least.

What this means for mortgage professionals

For those working in the mortgage industry, the debate matters in two distinct ways.

The first is rate trajectory. If Hunstad and Warsh are broadly correct - and the productivity benefits of AI begin to flow through to the inflation data more quickly than the sceptics expect - the Fed may have the cover to cut rates further and faster than current market pricing implies. Futures markets currently reflect only around 50 basis points of cuts through the rest of 2026, which would leave the 30-year fixed rate above 6%. A genuine AI productivity dividend, showing up in unit labour costs and core inflation by mid-year, could shift that calculus. Most forecasters, however, including Wells Fargo and Morgan Stanley, expect rates to remain in the mid-6% range through 2026 and into 2027. The scenario in which AI delivers a near-term disinflationary shock large enough to move monetary policy meaningfully is not the consensus view.

The second dimension is more immediate and more within the industry's control. Whatever the Fed does, AI is already demonstrably reducing the cost of producing a mortgage. Freddie Mac's 2025 analysis found that lenders fully deploying its AI-enabled Loan Product Advisor tools save up to $1,700 per loan and experience 40% fewer defects. The average cost to originate a retail mortgage reached approximately $11,800 in the second quarter of 2025.

Lenders that have leaned hardest into automation - document processing, automated underwriting, AI-powered servicing analytics - are operating materially closer to $6,900 to $7,000 per loan. That gap is not a monetary policy question. It is an operational one.

In this sense, Hunstad's broader argument - that AI is a supply-side force that expands capacity and reduces costs - is already playing out in miniature inside the lending industry itself. The companies capturing the most value are, as PwC found across industries broadly, those using AI to reinvent their economics, not merely to automate existing workflows. That distinction - between AI as a chatbot layer and AI as a fundamental restructuring of how a loan is produced and serviced - is likely to define competitive advantage in the industry over the next several years, regardless of what the Fed does.

Hold steady and watch

Hunstad's prescription for the Fed is deliberate patience. "I think the right course of action is to hold steady and put a communication out to the market that, 'hey, we're just going to hold steady and see what the true AI productivity benefits entail'," he said. The instinct is sound, even if the timeline is uncertain - and even if the Iran conflict and its attendant energy-price pressures complicate the picture considerably in the near term.

The larger question - whether AI will ultimately prove to be the disinflationary force that Northern Trust and the incoming Fed chair believe it to be - will not be settled in a single earnings season or a single monetary policy meeting. The 1990s analogy suggests the answer will arrive slowly, unevenly and only in retrospect with full clarity.

What is clear is that the argument is no longer a fringe position. When the head of a $1.4 trillion asset manager and the likely next chair of the Federal Reserve are making versions of the same case, the hypothesis that AI is structurally changing what the economy can produce - and at what price - deserves serious attention from anyone whose business depends on the cost of borrowing money.

The rate on a 30-year mortgage is 6.30% this week. Whether AI eventually moves that number, or whether the Fed does it first, or whether the two forces act together - that is the question that will define the next chapter of American housing finance. For now, the honest answer is that nobody knows. The smartest thing the industry can do is take both the potential and the uncertainty seriously - and not wait for either the data or the rates to move before acting on the efficiency gains already available.