The American starter home — the ranch-style two- or three-bedroom once relied upon by young families building wealth — is in trouble. The U.S. is short nearly 4 million homes, according to Goldman Sachs. A combination of soaring prices and restrictive zoning has squeezed out that first rung on the property ladder. For Vishal Garg, CEO of Better.com, the root problem lies deeper: the mortgage itself.
“One of the core reasons why people, even home builders, do not build starter homes in America is that nobody is willing to give mortgages for starter homes,” Garg said.
His argument hinges on how loan officers are paid. Mortgage brokers typically earn a commission of 1% to 2% of the total loan value. That means a much larger payout on a $1 million home than on a $100,000 unit — creating a structural incentive to chase bigger loans.
“American consumers with a mortgage amount under $300,000 are not treated well at all,” Garg said.
The result is visible in the numbers. The median first-time homebuyer age reached a record 40 last year, while first-time buyers now account for a record-low one-fifth of all purchases, according to the National Association of Realtors. Starter homes, usually under 1,400 square feet, are meant to be cheaper by virtue of their size. But shrinkflation has hit the housing market: new builds now cost 74% more and measure 11% less than a decade ago, according to Lendingtree. Median home size has fallen from a peak of 2,466 square feet in 2015.
Many younger buyers have effectively been priced out. Some turn to family for down payment help. Others are waiting on the sidelines for rates to drop below 6% and for inventory to improve.
Can AI close the affordability gap?
Garg is wagering that his company’s AI tool, Betsy, can make servicing smaller mortgages commercially viable. Better.com claims the system reduces loan processing costs from the industry average of nearly $12,000 — a figure cited by Freddie Mac — to $3,000. That $9,000 cut, Garg argues, makes it worth a lender’s time to serve starter-home buyers.
Garg also sees the tool providing underserved borrowers with advice typically available only to high-net-worth clients. Human brokers handling small loans often lack the incentive to offer detailed financial guidance. The AI, by contrast, functions as a research engine that can walk a borrower through steps to improve their credit profile — paying down a specific card, reducing a car payment — in order to reach a higher tier and a better rate. Poor credit accounts for nearly half of all mortgage denials for loans under $100,000, according to the Department of Housing and Urban Development.
When asked whether AI could solve the housing affordability crisis, Garg was blunt: “Totally.”
Why starter homes keep disappearing
Even as new builds shrink in size, higher price tags are pushing them beyond the reach of first-time buyers who need loans under $100,000. A 2022 Urban Institute report found that just 35% of home sales below that threshold were financed with a mortgage. Builders cannot construct cheaply enough to generate the loan amounts the system will actually fund.
Garg points to the mortgage process as a central culprit. Other analysts place more weight on zoning restrictions and builder economics. After the Great Recession, many developers shifted toward larger homes with higher margins, finding it easier to operate in markets designed around large lots.
Dennis Shea, a housing expert at the Bipartisan Policy Center, pointed to zoning as a key constraint in a recent interview with the Washington Post. “Home builders, particularly in the wake of the Great Recession, find it easier to build larger homes that have higher profit margins.”
For Garg, streamlining the paperwork side of borrowing is still part of the solution. Cutting mortgage costs, he argues, gives lower-income and first-time buyers a genuine opportunity to reach the property ladder.
“You are not saving $9 on a sweater,” he said. “You are literally saving $9,000 on a mortgage because of the combination of AI and machine learning.”