AI-Powered Mortgage Refinancing Could Reshape $9 Trillion Bond Market

AI slashes refinancing timelines to minutes

Mortgage lenders are deploying artificial intelligence systems that can approve refinancing applications in as little as two minutes, according to Bloomberg, a dramatic acceleration from the traditionally slow and paperwork-heavy process that has historically discouraged homeowners from securing lower rates.

Rocket Mortgage reports it can take borrowers from application to rate lock in 30 minutes and aims to reduce that to 10 minutes. United Wholesale Mortgage claims initial approvals in 15 minutes, while Better.com says it can complete the process in just two minutes.

The speed gains stem from AI systems that can instantly assess borrower eligibility, pull data from tax forms and bank statements, and determine optimal outreach strategies. Rocket has invested over $500 million in AI and automation technology over six years, CEO Varun Krishna told investors in May.

Doubling refinance participation rates

Currently, only about one-third of homeowners who could save substantially by refinancing actually do so, according to Morgan Stanley research. Many don't realize they're eligible or avoid the notoriously tedious process.

Morgan Stanley strategists, including Jay Bacow, co-head of securitized products research, project that AI-driven efficiency could double participation rates to approximately 60%. This would fundamentally alter the economics of 30-year mortgages, making them behave more like floating-rate instruments that adjust downward when rates fall.

Why it matters

This shift poses significant risks for investors holding mortgage-backed securities in the $9 trillion mortgage bond market. When homeowners refinance faster than expected, bondholders lose their stream of higher-interest payments sooner and must reinvest proceeds at lower prevailing rates. The result could be compressed returns and potential repricing of mortgage risk, possibly adding one or two tenths of a percentage point to home loan costs as investors demand compensation for increased prepayment uncertainty.

Labor costs and regulatory constraints

Labor represents roughly two-thirds of mortgage origination costs, according to the Morgan Stanley analysis. If AI substantially replaces human labor, closing costs could approach zero, creating what analysts describe as a "near-frictionless" refinancing process.

However, regulations still require licensed human participation rather than pure AI decision-making, and legally mandated document review periods add unavoidable delays. Upfront fees for appraisals, legal work, and mortgage taxes also remain barriers to refinancing regardless of approval speed.

Market skepticism remains

Not all industry participants believe AI will transform refinancing behavior. Jeffrey Kvalevog, chief strategy officer at New American Funding, noted the industry has seen similar promises about optical character recognition and automated communications that failed to revolutionize the business.

Mark Tecotzky, co-head of credit strategies at Ellington Management Group, said existing prepayment data shows evolutionary rather than revolutionary change. The real test will come during the next significant rate decline, when refinancing demand surges.

"We haven't had a good refinancing wave since AI became a thing," said Walt Schmidt, a mortgage strategist at FHN Financial, noting that current rates above 6.5% have suppressed refinancing activity.

These details were first reported by Bloomberg writers Trapanick and Carpenter.