Essay

Opendoor fixed the buy button. Last quarter, the houses it sold just paid the interest on the houses it holds.

Opendoor made $34 million on the 2,728 houses it sold last quarter, before interest. The interest on the money it borrowed to hold the rest came to $34 million. That pairing is mine, two lines from the company’s own earnings release, and it is the honest way to grade an iBuyer. The business is a market maker. It buys your house now so you can skip the wait for a buyer, and it earns a spread for doing the waiting itself. The spread on what it sells has to pay for everything it still holds. A year ago it fell short, $20 million against $23 million. In the spring it paid twice over. This fall it paid exactly.

A disclosure first. I worked in growth engineering at Opendoor from December 2024 until January 2026.

The buy side works. Opendoor bought 4,865 houses in the quarter, against 1,169 a year earlier, when it had pulled way back, and it spent far less to win each one. CEO Kaz Nejatian credits what he calls “Broader product choices, less friction and better customer education”. Grant all of it. The offer got cheaper to make, and more sellers said yes.

Selling is where the wait lives. A buyer still has to find the house, get a mortgage and close, and a pricing model can only shorten that by giving up margin. So the houses piled up: 7,596 of them at the end of September, carried at $2.57 billion, mostly on borrowed money. The lenders know exactly what they are financing. Opendoor’s 10-Q says the amount it can borrow against its homes can shrink “as properties age beyond certain thresholds.” The clock is written into the loan. After the quarter, Opendoor signed new credit lines running to 2030, which buys time on the debt and none on the houses. Its guidance for the fourth quarter shows the trade it is making now, more revenue and less profit, in one plain sentence: “Our focus is moving homes.”

Then the market moved, on houses already bought. Freddie Mac’s 30-year mortgage rate went from about 6.5 percent in late June to 7.3 percent in the last week of September. Opendoor’s guidance concedes the pace at which homes go under contract “can change quickly, as they have in the last few months.”

Opendoor’s best defense is a good one. The houses it sold this summer made more money than last summer’s. Its lenders just extended it to 2030. Fewer of its listings had gone stale than the market’s: 18 percent had sat four months unsold, against 30 percent by the company’s own count of comparable homes. And Cash Now, More Later, where the seller takes part of the price up front and the rest when Opendoor resells, now brings in more houses than all of Opendoor’s products did a year ago. That hands some of the wait back to the seller. On that reading, the third quarter is a market maker caught by a September rate jump.

The answer is what Opendoor promised in August, with the slow season already priced in. Its second-quarter release told investors to expect a margin on homes sold of “around 4% to 4.5%,” and said the usual fall slump would be “significantly narrower, reflecting the structural health of our underlying cohorts.” The slump was narrower. The margin still came in at 3.3. Now, on its profitability goal: “We are currently behind but we intend to catch up,” because “a slower market has put us behind our goal.” The admission is worth more than the alibi. Rates did jump, and that explains part of the miss. The season was already in the guide.

The stale listings say the same thing. Their share doubled in the quarter, from 9 percent to 18, starting from a freshly cleaned-out spring book, and in a quarter when all that new buying should have pushed the number down: nothing bought after July 1 could have sat four months by September 30. It doubled anyway.

Offerpad made the same bet on a thinner cushion. It bought 605 homes and sold 210, nearly three for one, with $12 million of cash in the bank. CEO Brian Bair blames mortgage rates, which in his telling “rose more than 70 basis points from mid-August to early October,” and Freddie Mac bears him out. The release leads with its fee channels, where it hands the seller to other buyers or simply lists the house, reaching 35 percent of closings. That share rose mostly because its own sales fell 43 percent. Then, per its 10-Q: “we began slowing our home acquisition pace in October and expect to continue doing so for the foreseeable future.”

What I took from my years in title and escrow, long before Opendoor, is to route around the slow transaction and never bet on it speeding up, and my own later bets left the real estate transaction altogether. Cash Now, More Later is the closest either company has come to a route around. Everything else on both balance sheets is carrying, which is the job, and the third quarter priced it. The buy button works. The sale pays for every house the button bought, and this fall it paid with nothing left over.

Sources: