Uber Technologies, Inc. UBER

Key data

The business

Uber runs three marketplaces off one app. Mobility (ride-hailing) is the profit engine, Delivery (Uber Eats) is the scale engine, and Freight is a rounding error it keeps mostly for optionality. In each, Uber owns neither the car nor the kitchen. It owns the demand, matches it to supply, and takes a slice of gross bookings, roughly a fifth to a third depending on the segment. That is the whole moat: the rider opens Uber because that is where the drivers are, and the driver signs on because that is where the riders are, a two-sided pull that a new entrant has to buy both sides of at once. The texture worth knowing: Uber does not even own its own map. It licenses Google Maps under a master agreement it just amended a fourth time, a reminder that some of the plumbing under the toll booth belongs to somebody else.

The numbers

Read operating income, not the bottom line. Reported net income is whipped around by marks on Uber's equity stakes (Aurora, Didi, Grab) and a one-time tax-asset release that inflated the third quarter of 2025. The operating line is the clean read.

The net-income column swings from $6.6B to $0.3B to $2.4B in three quarters on almost no change in the business. That is the equity marks talking, not the marketplace. Note too that Q2 2026 was the second-biggest revenue quarter Uber has ever posted, behind Q4 2025's $14.4B, not the record.

Now the clean signal across the cycle.

Operating income doubled from 2024 to 2025 and ran another 42% ahead in the first half, from $2.7B to $3.8B, while revenue grew 13%. Profit growing three to four times faster than sales is the operating leverage of a marketplace past its fixed-cost hump: each incremental trip drops most of its take to the line. Operating cash flow reached $10.1B in 2025 against nominal capital spending, a 6.6% yield on today's price. Against the two consensus prints that exist, both far out, FY2028 earnings of $5.52 put the stock at under 14 times, and the operating trend is running ahead of the pace that number assumes.

They are spending the cash two ways.

Buybacks went from a gesture to $6.5B, then $4.8B in half a year, shrinking the count while the multiple is low, the right time to do it. The debt jump is new: Uber drew fresh term loans to fund a debt-financed acquisition, papered in the August credit agreement. So cash fell from $7.1B to $4.9B even as the business threw off more of it. The variant here is plain: the market prices Uber at the 10th percentile of its own decade on sales because it fears autonomy and a tapped-out take-rate, and the read is that operating income keeps compounding through both. The print that settles it is gross-profit growth staying ahead of revenue growth.

Management

Dara Khosrowshahi's record is now written in the capital account, because insiders neither bought nor sold in the open market over the past year. The tell is the buyback turning on hard at a low multiple rather than at the 2021 highs, and debt raised to buy a competitor rather than to plug a hole. Both are owner-minded moves. The watch-item is the one every acquirer earns: $2.2B of new borrowing and an integration to prove out, on top of a stock-comp bill still running about 3.5% of revenue that the buybacks are partly there to mop up. Pay detail is not in this run's filings and is left out rather than guessed.

How it fails or surprises you

Autonomy routes through the app (right tail). If Waymo and the fleets that follow it plug into Uber for demand, Uber sheds its biggest cost, the driver's cut, and keeps the rider. That converts the great fear into a margin windfall. Nobody is paying for it at 2.8x sales. The first print is autonomous trips on the platform growing as a disclosed line.

The toll booth gets walked around. Regulators reclassify drivers as employees, or drivers and restaurants push back on the take, and the slice thins. Operating income eased from $1.92B to $1.89B in the latest quarter, a soft print but one quarter alone does not break the thesis. If gross-profit growth falls behind revenue growth for two quarters while autonomous volume builds outside the marketplace instead of inside it, the toll booth is being bypassed and the moat is gone.

The number you trust least is the reported one. Net income of $0.3B in two of the last four quarters says the equity portfolio can swamp the P&L. A hard markdown of Didi, Aurora or Grab would crater EPS and book value while operations compound underneath, and if the operating read is wrong, the marks are where you would see it first.

Closing thoughts

The market has already discounted the scary story: 2.8x sales and the low end of a ten-year range is not what a business gets when investors expect it to keep compounding, it is what they pay when they think autonomy ends it. So the edge is not information, it is weighting. On the other side of the position sit people extrapolating the AV threat straight through the take-rate. The read beats theirs only if operating income keeps outrunning revenue, which it has done every quarter of this cycle. The fatter tail is up, because a cheap multiple and a $4.8B buyback cushion the downside while the autonomy option is free, but the left tail is real and specific: if the app becomes a dumb pipe the AV owners bypass, the moat is gone and no multiple saves it.

The bet is still that Uber stays the place people reach for when they want a ride or want dinner, in more cities every year, keeping its cut of each trip, and the self-driving cars show up inside Uber's app hunting for riders instead of going around it. What breaks it is two numbers moving wrong together: gross-profit growth falling behind revenue growth while autonomous volume builds outside Uber's marketplace instead of inside it. Watch those two for two quarters. If they diverge, the toll booth is being walked around and the read is wrong, whatever the buyback is doing to the share count.