Rocket Companies, Inc. RKT

Key data

The business

Rocket makes home loans over the phone and the web instead of through a branch, and it is the biggest retail mortgage lender in the country. It does not keep most of those loans. It sells them to Fannie Mae and Freddie Mac and keeps the right to collect the monthly payment for a fee, which is the servicing business. Two things happened in the second half of 2025 that changed the shape of the company. In July it bought Redfin, the home-search site and brokerage, for stock. In October it bought Mr. Cooper, the largest mortgage servicer in the United States, again mostly for stock. So the company that used to live and die on how many loans it could write in a quarter now also collects payments on a vast book of other people's mortgages and owns the front door where buyers start looking. The moat is the brand and the recapture machine: when rates fall, Rocket already has the borrower's file and can refinance them before a competitor calls. That is worth more now that the servicing book is several times larger.

The numbers

The story is a company climbing out of a trough while its own shape shifts underneath it. Start with the last five quarters.

The inflection is Q4 2025, the first quarter with the acquired businesses inside it, and the two prints since are the real signal: profitable, at scale, on more than $2.6B of quarterly revenue against $1.5B a year earlier. Read the last two lines with care, though. Revenue fell 3.5% from Q1 to Q2 and net income fell about 23%, so the recovery is not a straight line up, and the reorganization that closed June 30, 2025 makes the per-share figures before and after it different animals.

The 1H 2026 row is the whole argument in one line: the first half of this year already earned more than any full year since the 2021 refinancing boom. That is the acquisitions folding in and rates easing off their peak at the same time. The serviced balance that more than tripled a year ago is now the Mr. Cooper book carried in full, and flat quarterly revenue against that far larger platform is the gain-on-sale squeeze on new loans still bleeding through even as servicing scales, so the prior year's watch-item held rather than reversed. Here is the variant: the market is treating the recovery as fragile because origination is still weak, but the earnings now come from a servicing book that pays whether or not anyone refinances. What settles it is one clean quarter where servicing income, not a volume spike, carries the profit.

Management

The record is mixed and expensive. Diluted shares are up 25.5% in three years, the price of buying Mr. Cooper and Redfin with paper and collapsing the old ownership structure, and there have been no buybacks since 2022. Free cash flow ran roughly −$4.0B in 2025, heavy with the reorganization and deal costs. Against that, CEO Varun Krishna was paid $52.9 million in 2025, double his $25.9 million the year before, in a year the company reported a small GAAP loss. Insider selling is noise, 42 small open-market sales totaling about $2.0 million with no buys, which reads as routine tax withholding and comes to a hundredth of a percent of the company; founder Dan Gilbert still controls it through Rock Holdings, so the economics that matter never hit the open market. The honest read: the team made a large, forward bet with shareholders' equity and paid itself well while doing it.

How it fails or surprises you

Rates fall and the recapture machine fires (right tail). With the largest servicing book in the country now in hand, a real drop in mortgage rates hands Rocket a list of millions of its own borrowers to refinance before anyone else reaches them. The market is paying for today's weak origination, not this. The first Fed cut that pulls 30-year rates through a refinance threshold shows up as a jump in locked volume within a quarter.

They bought the top of the cycle. If rates stay high into 2027, Rocket paid up for a servicing book and a brokerage while its core origination stays starved, having diluted holders a quarter to do it. The tell is gain-on-sale margin and origination volume both staying flat for two more quarters.

The profit is thinner than the price. GAAP earnings barely cleared zero last year while the stock trades at 90 times trailing earnings and 1.7 times book, so the read that the recovery is real gets disproved if 2H 2026 net income slides back toward breakeven.

Closing thoughts

This is two questions stacked. One a print resolves: does the giant servicing book convert into recapture revenue when rates fall, which the first easing cycle answers directly. The other no print resolves: the mortgage cycle itself, where the only defense is surviving the years when nobody buys or refinances, and Rocket, with its debt load and thin GAAP earnings, has less cushion than its market cap implies. The right tail is fatter than it looks because the servicing book pays in the meantime, but the left tail is a full cycle of high rates grinding down a company that spent its balance sheet on scale. What is at risk if origination stays dead is real; what the recapture is worth if rates break is larger.

The bet, still, is that Rocket stays the first place people go for a home loan and now keeps them through the servicing book long after, walking search to loan to payment and back. What breaks it is rates that never fall far enough to make the recapture pay. The pair to watch is gain-on-sale margin on new loans and the recapture rate on the serviced book after the first rate cut; if the second does not climb when the first turns, the machine everyone paid for is not there.