Summary: a compounder?
Uber’s third quarter looks like more of the same: bookings up 20% in constant currency, the fifth straight quarter at that pace. Underneath, the engine has changed. Trips slowed to 16%, more of the growth came from each trip being worth more, and Mobility earned its best margin while keeping less of each ride. Meanwhile the autonomy race is only now leaving the grid.
By the end of 2026 Uber Technologies, Inc. expects about 700 driverless cars on its network, over 2,000 by the end of 2027, then what it calls a much larger commercial ramp from 2028. It is a small field so far, more than half of it looks to be one partner’s cars, and Uber has met fewer than half of its firm dates on time on the firm’s count.
All of which feeds one question. Has Uber become a longer-term compounder rather than a high-growth stock? The firm asks three things of a compounder, and Uber’s answers are uneven.
| Test | The answer so far | What would change it |
|---|---|---|
| Can it keep finding room to grow? | Yes, though part of the room is being bought: Delivery Hero adds about 22% to bookings |
Bookings growth below about 13%, where the price sits on the firm’s grid |
| Does it generate cash? | Yes. Owner cash, free cash flow less stock pay, is $3.18 a share over twelve months once the insurance float is out, and $4.08 with it |
The margin without the float falling from 2.9% |
| Is the cash being spent well? | Not proven: the largest use is Delivery Hero, which pays only if most of Uber’s synergies arrive (section 6), and buybacks have paused for it | The offer, the close, the synergies by 2029, and buybacks back to most of owner cash, as in 2025 |
- Share price, 5 November 2026
- $68.78
- Model value on the firm's 20% growth case, without the float
- about $89, firm's model
- Rides a day for a robotaxi to match a driver, on the firm's sum; Waymo's cars do 19
- 20 to 24
Read through the firm’s model, the price is not stranded. At today’s 20% bookings growth, fading to 5% over ten years, the model gives about $111 a share on the reported margin and about $89 on the margin without the float; at trips’ 15.6%, about $76 (the firm’s calculation; section 8 states the inputs). The firm’s own case is that 20%, the middle of Uber’s guide for Q4, so the $89 is the firm’s number, an implied return of about 13% a year from $68.78 if the model’s cash flows arrive. So the shares are a bet on Uber holding bookings growth near 20% and on its cash being spent well, with no help needed from robots before 2028. The firm takes Uber’s $1.2bn of synergies as its own expectation, for the reasons section 6 gives, and expects buybacks to resume once the deal closes. The table above says what would change its mind, and each has its date: growth and margin are read each quarter, the offer’s result is due in November, the close in the second half of 2027, the synergies by 2029, and buybacks from the first full quarter after close.
Terms in this note
Bookings: the total value of rides and orders, including taxes and tolls, before consumer discounts and driver pay. Owner cash: free cash flow less stock pay. The float: the cash that builds up as insurance reserves grow, owed to claimants in time. Take: what Uber keeps of a booking. Constant currency: growth with exchange rates held still.
The shape of the business
Revenue is the measure only here, because Uber changed how it books UK rides in January 2026, which takes about $1.1bn a quarter out of Mobility revenue with no change in the business. The charts add it back so the period reads on one basis, which is the firm’s calculation.
Share of revenue by segment, 2022 to Q3 2026, with the UK change added back (firm's calculation). Source: Uber releases.
Delivery and Freight have been taking share from Mobility since the end of 2024, though Mobility is still more than half. The whole company has moved from a loss to 15.0% of revenue on adjusted EBITDA less stock pay, the firm’s measure, which takes off what owner cash takes off. Uber still discloses adjusted EBITDA for the group but no longer calls it a key measure; it now leads with non-GAAP operating income and earnings per share.
Adjusted EBITDA less stock pay, by quarter and as a share of revenue, with the UK change added back (firm's calculation). The two scales are set so their zeros align.
From here the note argues on measures accounting does not touch: users, trips per user, bookings per trip, profit per trip and owner cash. These are the firm’s dials, and section 10 sets them out for later notes to mark.
Volume has slowed, the headline hasn’t
The market’s fear is that growth has peaked, and the headline does not support it. In constant currency, bookings grew 22% in Q1 2023 and 20% in Q3 2026. The parts tell a different story.
What bookings growth is made of, in constant currency, year on year; the split is the firm's calculation, good to about half a point.
Trips grew 25.0% in Q3 2023 and 15.6% in Q3 2026. Trips per user, the number that says people are making Uber a habit, slowed from 9.1% to 1.6%. Users held up for longer, then slowed from 18.1% to 13.8% in three quarters. What kept the headline at 20% is bookings per trip, which in Mobility and Delivery turned from about −1.6% to about +3.4% (the firm’s calculation).
The CFO puts that rise down to mix, more US, Delivery, premium and grocery, “rather than higher like-for-like pricing”. That is management’s account and the firm has not tested it. It matters because mix can only do so much for so long: a business growing on bigger baskets and longer trips is a different one from a business growing on more people riding more often. About half of Q3’s slowing against Q2, on the CFO’s account, is Trendyol GO, bought in June 2025, entering the base.
Where the room is, on Uber’s own figures: about 55m Uber One members place about 70% of Delivery bookings; grocery and retail run at $15bn a year, up 40%; premium rides are up 35% and business travel about 50% (the first two from the call’s machine transcript). None of it is new volume on the dials, which is the point for the first test: the next leg of growth is more spent per user, not more users.
The quarter Mobility slowed and earned more
Mobility bookings growth fell to 17% in Q3, and its margin on bookings rose to 8.0%, the highest in the period. Uber’s explanation is drivers: supply that “has not kept pace” with demand in several international markets, with fuel costs from the Middle East war in Brazil and India, immigration in Canada, and some pressure in the US. Its response is to move incentives from riders to drivers. Supply hours accelerated in October, the CFO said on the call, without giving a number.
Management is explaining its own miss, and one thing does not fit. Monthly users across the platform, the only count Uber gives, slowed too, from 17% in Q1 to 14% in Q3, back to where they ran before Trendyol GO lifted them in 2025; a shortage of drivers does not obviously explain even that.
Mobility bookings growth in constant currency against segment margin on bookings. Two scales: where the lines cross means nothing, the opposite directions are the point.
Growth held at 19% to 20% until one quarter fell to 17%, while margin was already rising, so slowing did not cause the gain. To see what did, follow a $20 ride.
What Uber keeps of a $20 Mobility ride after paying drivers and consumer incentives, split into other costs and segment profit. The $20 ride reads the percentages; it is not Uber's average fare.
Uber keeps less of each ride, 33c less in six quarters on this measure, and its other costs fell by more, 43c. So profit on the ride rose 10c while what Uber keeps fell: running costs fell faster. Lower insurance is the filed reason, down $41m while Mobility bookings grew 17.5%. Uber does not disclose its level, so the firm cannot say how much of the 22c fall in other costs over the past year it explains.
The year-on-year fall in what Uber keeps narrowed from 26c to 11c, and the CFO said on the call that Q4’s supply spending will hit take rate rather than margin. Uber’s translated guide for adjusted EBITDA was in line with consensus while its earnings guide was about 8% below LSEG’s consensus on the firm’s sum (via Reuters), and the cause of that gap is not yet clear. Q4 is the test, and the margin it tests is the one the second test rests on.
Robots: small now, and who pays
Autonomy cannot fix the driver shortage for years. What it can do, and what it costs, depends on how busy a car is and who owns it. On the firm’s sum a car needs about 24 rides a day to earn Uber what a driver’s rides do, and Waymo’s cars do 19.
A small field, with a long ramp
Uber's stated driverless fleet, one dot to 100 vehicles. Percentages are the firm's estimate of each fleet's capacity against 320m Uber rides and deliveries in one peak week.
Uber is not short of cars in principle: the CEO said at a September conference that it has secured access to over 120,000. But that number has no date, and the plan does. Even 10,000 cars would be under 1% of a week’s trips, so human supply is the whole answer until at least 2028, and the shortage is mostly in Brazil, India and Canada, which is not where the first fleets are going.
The record so far
The firm went through every promise it could find that Uber has made on autonomy since 2023, 101 in all, and checked what happened. This is the firm’s own scoring, of the firm ride promises due by 6 November 2026, counted once each.
Uber's firm autonomy promises due by 6 November 2026, by the date promised, against what the firm could find. The scoring is the firm's.
Six came on time, five late, and for five the firm found nothing: 6 of 12 on time before 2026, 0 of 4 in 2026, too few to call a trend. Uber gives no reason for any of the slips. In the Bay Area both the date and the form moved: a launch in 2026 became a driverless launch in Q1 2027. A heavier test is close, as 11 more fall due by 31 December on the same count, eight launches and three fleet or city counts, the 700 cars among them.
Who owns the car
Three answers appear in Uber’s deals: Uber, a fleet partner, or the AV company. Uber’s Waymo cars are Waymo’s, about 400 of the 700, more than half of the year-end fleet, with a partner whose exclusivity with Uber ends in January 2028. On the call the CEO said early cars go on Uber’s balance sheet because lenders cannot yet price what they will be worth, and the CFO that Uber wants third-party capital from the 10,000 level.
Vehicle spend does not look like the constraint. Uber’s 10,000+ for 2028 and beyond, a floor with no year-end date, at $100,000 a car, a Lucid Gravity at list plus a sensor kit at the cost press reports attribute to Waymo, about $20,000, would be $1.0bn if Uber owned every one, and about $200m at the one-fifth owned share the firm’s model assumes, on this section’s $100,000 a car, against $6.5bn of owner cash without the float in the last year; the $50,000–60,000 car the CEO said in September that partners can see would halve it. Vehicles are the small end of the bill: the CEO said on the call that Uber has about $10bn of commitments across partners and fleets, subject to milestones, and the 10-Q now files vehicle offtake commitments without figures and $1.3bn of exposure to fleet operators, a $449m loan to Moove among it. Management ranks autonomy ahead of buybacks.
The sum below is Uber’s own cars. The larger question is rides that never reach Uber: Waymo does over 400,000 a week on about 3,000 cars, most of them on its own app, against about 400 cars on Uber, and Tesla runs in six cities on its own report. Every tenth of Mobility volume that goes that way is about 46c a share of profit before tax on today’s margin (the firm’s sum), which the grid in section 8 capitalises for ten years. Uber’s answer is that new supply makes new trips: in mature AV markets, on its figures, rideshare grew 7 points faster than the rest of the country.
Does a robotaxi pay?
Nobody outside these deals knows, and no Uber fee, take rate or robotaxi insurance premium is public. So the firm built one sum for one Uber-owned car in the US on a $20 ride, and sourced each of its eleven lines as far as a source exists; the fold below names them. It covers the US only, because half of the 2,000 cars are abroad, where the firm’s cost data is too thin to model. A robotaxi need not bring more rides. The same rides move from a driver to a car, the day’s early rides pay the fixed costs, and only the rest leave Uber better off than a driver would.
One car on one day, on the firm's sourced sum, set out in the fold below. Waymo's cars do 19 rides a day on its own figures; Uber claims high 20s.
On these numbers the fixed costs clear at about 20 rides a day, and a car needs about 24 to earn what Uber earns on the same rides with a driver. That is above what anyone has shown. Waymo’s whole fleet does 19 a day, over 400,000 rides a week on 3,000 cars on its own figures; WeRide reports over 21 on average and 28 on its best day, and Pony.ai 23, both at China’s fares and costs. Uber’s own claim of rides in the high 20s rests on the unofficial transcript of its call and on a few hundred cars in markets where drivers take the peaks, and in writing Uber claims only a relative advantage, about 30% more trips per car than first-party fleets. At Waymo’s 19 a car loses about $12,000 a year against a driver’s same rides; at Uber’s 28 it earns about $12,000 more (the firm’s sum).
What the sum rests on
The sum’s spread is the sourcing, not the arithmetic. Each line has a sourced end that helps the car and one that hurts it, and with every line at the helping end a car matches a driver at about 12 rides a day, with every line at the hurting end at about 55; the three lines that move it most are constructions, named in the fold below.
Insurance is the line that decides it. The sum charges the robotaxi what a driver’s car costs Uber in a middle state, 21% of the fare, on Uber’s own state figures, which run from 13% in Colorado to 33% in upstate New York; in Texas, where the first fleets run, it is 16%, about $0.38 a mile, and the break-even about 21. A robotaxi rate half the human one, which Swiss Re’s claims data for Waymo would support, would bring the break-even to about 20, level with Waymo’s 19, and so would the CEO’s $60,000 car. So on what is public the firm reads a robotaxi as at or just short of paying its way, one to two rides a day on the lines that run where the fleets do, and the lines that would settle it as the ones Uber and its partners have not disclosed.
It is a test of the model more than a change to the earnings: if Uber owned all 10,000 cars of its 2028 floor and every one did 28 rides, they would add about 6c a share a year before tax, and at Waymo’s 19 they would take about 6c away (the firm’s sum). A car that is not busy favours Uber not owning it. If a partner owns the car and Uber takes a cut of the fare, owning wins only above about 28 rides a day against a 15% cut, and about 33 against 20%, on the same sum. The same sum says who can own the car: a fleet partner paying that cut loses about $3 a ride at Waymo’s 19 and breaks even near Uber’s 28, while a developer running its own system clears it at either (the firm’s sum). So the question is what take Uber can hold once others own the fleet, and at today’s utilisation the owner that can afford it is the one that can leave.
The robotaxi sum, line by line
| Line | Per ride, 19 rides | Per ride, 28 rides | Per car, a year at 28 |
|---|---|---|---|
| Fare | $20.00 | $20.00 | $179,200 |
| Taxes and tolls | −$2.00 | −$2.00 | −$17,920 |
| Driving system fee | −$6.81 | −$5.49 | −$49,192 |
| Card fees and support | −$0.52 | −$0.52 | −$4,659 |
| Insurance | −$3.33 | −$3.33 | −$29,882 |
| Charging and upkeep | −$1.33 | −$1.33 | −$11,953 |
| Depot and remote assistance | −$2.30 | −$1.56 | −$14,000 |
| Vehicle | −$4.11 | −$2.79 | −$25,000 |
| Profit | −$0.42 | $2.97 | $26,595 |
| Ride with a driver | $1.60 | $1.60 | $14,336 |
Fixed costs are about $64,000 a car a year and each ride adds $8.51 over a ride with a driver, so a driver is matched at 23.5 rides on 320 days and 6.67 miles a trip, the miles from California’s regulator. The per-car column is at Uber’s 28. Every line is the firm’s sum on sourced parts, and the sources are named in the fold at the end of the note. Three lines move the answer most, and all three are constructions, because nobody publishes them: the fee a driving system’s owner charges, $40,000 flat to $72,000 on WeRide’s and May Mobility’s figures, from 22 to 31 rides a day; the fare and the miles it buys, as a $25 fare over 10 miles reads very differently from $20 over 6.7; and what a depot and remote assistance cost, $9,500 to $30,000 a car on Waymo’s and Cruise’s staffing. What each line rests on:
- Fare: between the US median and mean rider price in a vendor survey; a reading device.
- Taxes and tolls: 10%, a low-tax city; New York’s filed charges are 21%.
- Driving system, a $25,000 floor plus 15%: the form from WeRide’s filings, the size from WeRide and May Mobility, $40,000 to $72,000 a car a year.
- Card fees and support: Uber’s filed card costs and operations and support, as shares of bookings.
- Insurance, $0.50 a mile: what a driver’s car costs Uber in a middle state, 21% of the fare on Uber’s own state figures, 13% to 33%, over a vendor’s miles. Uber’s filed reserves put the cost at 5% of Mobility bookings worldwide but give no rate per mile, and no robotaxi premium is public.
- Charging, maintenance and cleaning, $0.20 a mile: power, consumption and maintenance from the EIA, the EPA and the AAA; cleaning unpriced.
- Depot and remote assistance: built up from Waymo’s and Cruise’s staffing ratios, $9,500 to $30,000 a car.
- Vehicle, $100,000 over four years: Lucid Gravity at list plus a sensor kit at the cost press reports attribute to Waymo; life from Pony.ai’s filings.
- Ride with a driver, Q3 2026: Uber’s filed Mobility margin, worldwide, set on a US fare.
New markets: what Delivery Hero buys
In July Uber agreed to pay €41.50 a share in cash for Delivery Hero, the Berlin food-delivery group behind talabat in the Gulf and Baemin in South Korea: about $14.8bn for the whole company, $13.7bn after the stake it already held. Uber keeps 50 of its markets; Delivery Hero sells 14 to SSW, a private buyer Uber is lending most of the price. It is the largest answer to whether Uber can keep finding room, and it takes the cash a compounder would return.
What arrives, and when
Uber's 2025 bookings from its releases, plus the $42bn Uber attributes to the 50 markets it would keep; the total is about $236bn, as Uber states.
Nothing arrives until the deal closes, expected in the second half of 2027, and the firm has found no regulatory clearance yet. Uber’s stake and the undertaking of Prosus, which holds about 17%, cover 41% of the shares on the firm’s count, against a threshold of 50% plus one, and the offer’s result had not been published when this note was finalised, on 6 November 2026. Bookings would be about 22% larger. Delivery’s share of bookings jumps from 47% to 56%, though Delivery should pass Mobility during 2027 on its own, at today’s growth rates. The deal speeds up a shift already under way.
What it costs
| When | What happens | In Uber’s numbers |
|---|---|---|
| Paid already | $3.94bn for shares and swaps, a 24.5% stake |
No bookings |
| Since September 2026 | €4.5bn of euro notes at a 4.44% average coupon, about $231m a year |
Interest, stated for general purposes, with nothing consolidated |
| Second half of 2027 | Close, with about $9.8bn more cash; long stop November 2027, extendable to May 2028 |
Bookings start |
| Within 18 months of close | Over $1.2bn of synergies, on Uber’s figure |
Full effect by about 2029 |
Owner cash per share without the float, on today's figures with no growth (firm's rough sum, set out in the fold below). Synergies are Uber's $1.2bn, within 18 months of close.
On today’s figures, the deal lowers owner cash per share by about 12% at close, and it pays only through the synergies. Delivery Hero looks profitable, with about $339m of free cash flow on its 2026 guide, but that is before its interest, stock pay and the talabat minority, and Uber has about $577m a year of funding to cover on top. So it needs about $850m of synergies, 71% of Uber’s figure, just to beat buying back shares with the same $13.7bn. On company-compiled consensus for Delivery Hero’s 2028 cash flow, the figure falls to about 42%, and the fall at close to about 7%. Uber’s claim is that the deal adds to non-GAAP earnings per share at close and by a high single-digit percentage by year three. Both can be true, as that measure does not take off stock pay. If the offer lapses, Uber keeps its 24.5% stake, €4.5bn of notes raised for general purposes and the buyback it paused, and owes a €700m fee only if regulators are what stops it; on the firm’s sum that is no worse per share than the deal with half its synergies, so the position does not depend on the offer succeeding.
The owner cash sum, line by line
| Line, a year | $m | Status |
|---|---|---|
| Funding $13.7bn: euro notes, term loan, and $3.94bn already paid | −577 | Rates filed, mix estimated |
| Delivery Hero free cash flow, whole group, 2026 guide | +339 | Filed |
| 14 markets going to SSW | +89 | Estimated, on EBITDA only |
| Delivery Hero net interest | −272 | Company-compiled consensus, less a repricing saving |
| Delivery Hero stock pay | −288 | Filed |
| talabat minority’s dividends | −81 | Filed |
| At close | −790 | |
On $6,522m of owner cash without the float and 2,050.3m shares |
$2.80 a share |
Firm’s calculation |
Not included: tax on synergies, about −$0.09 a share at 15%; one-off costs to achieve them, for which no source exists, and deal costs of up to €240m; ownership below 100% at close; legal items Delivery Hero’s guide excludes, a net €655m in 2025. The kept markets’ own cash flow is not disclosed. The buyback comparison uses 199m shares at $68.78 and the same funding: $3.21.
Where Uber would be strong
Uber's or a kept Delivery Hero brand's share (*) against its largest rival. Dashed bands are the European Commission's 2024 ranges; rival dots sit at the middle of theirs.
In Europe most of what Delivery Hero leads goes elsewhere. The 14 markets going to SSW are named in the agreement: Austria, the Czech Republic, Norway and Sweden under foodora; Greece, Cyprus, Poland, Portugal, Romania, Moldova and Spain, mostly under Glovo; Chile and Ecuador; and Turkey, about $11bn of 2025 bookings between them. On the Commission’s 2024 ranges for 20 countries, Delivery Hero leads 8, and 7 of them are on that list (the scoring is the firm’s). Uber or a kept brand would lead in 4: Belgium, France, Hungary and Portugal. What Uber keeps is 50 markets with about $42bn of 2025 bookings on Uber’s figures, and Delivery Hero’s Asia and Middle East segments alone booked €35bn of its €49bn last year, so most of what is kept is South Korea, the Gulf and Asia, which is where Coupang, in Korea, and Meituan’s Keeta, in the Gulf, are attacking, and talabat’s margin fell from 6.0% to 4.8% between Q4 2025 and Q1 2026, before 5.0% in Q2. Uber gives no split, but that is where most of the synergies would have to come from.
Are the synergies believable?
Uber’s own Delivery is the base the synergies would land on, and it is pausing: bookings grew 23% in Q3 in constant currency, but the gain in what Uber keeps of each order faded from +96 basis points year on year in Q1 to +3 in Q3, the margin slipped 14 basis points on the quarter to 3.70%, and segment profit was flat at $1,057m on $1.1bn more bookings (the firm’s calculation from the 10-Q). Management describes reinvesting aggressively. Whatever 4.3% asks of the combined business below, it asks it of a segment that is choosing not to widen its margin today.
Adjusted EBITDA on bookings, before stock pay, latest year or annualised quarter. The dashed bar is the firm's test of Uber's $1.2bn of synergies, if all of it lands on Delivery.
Size does not decide the margin here: talabat earns most at a tenth of DoorDash’s scale. No company splits margin by acquired market, so nothing can prove a deal lifts it. What the filings do show is profit following local strength, and entrants spending to break it. The kept markets earned about 2.6% in 2025 on Uber’s figures, under IFRS, which leaves lease cost out; put it back on Uber’s basis, from Delivery Hero’s lease payments, and they earned about 2.3% (the firm’s calculation). On that basis Uber’s figure, taken at $1.2bn, would take combined Delivery from about 3.4% to about 4.3% (the firm’s calculation). That is above DoorDash and below talabat, which is not absurd, but it asks a $133bn business to earn close to what a small, locally dominant one earns. The firm takes Uber’s figure as its own expectation. Uber’s order puts the largest line on moving to one technology platform, then headcount, support and shared services such as payments and insurance, with cross-selling last and small: cost Uber controls, rather than revenue it has to win. The Berlin undertakings hold the headcount line back until 2029, so the risk the firm sees is timing more than size.
Home turf: the US and the Amazon deal
The US market is consolidating around the leader: on 29 October 2026 DoorDash agreed to buy Instacart, which would put about 77% of US delivery spending on one platform if regulators clear it (the firm’s sum on Consumer Edge’s shares).
US delivery spending, Consumer Edge via AP, stated 29 October 2026; the total includes Instacart. Uber Eats is just under 20; 77% is the firm's sum.
On 5 November 2026 Amazon made Uber Eats its Prime restaurant partner, replacing Grubhub: $0 delivery on orders of $25 or more, ordered on Amazon with no Uber account, in nearly 10,000 US cities and towns. Who funds it is not disclosed, and it matters, as share bought with Uber’s money shows up in the margin. The precedent is sobering. Grubhub’s share roughly halved in its years as the Prime partner, though it was also for sale and cutting marketing, so that is no clean test. The firm’s rough cases have Uber Eats gaining 0.4 to 3.6 points over two to three years, about 1.3 in the middle, worth about $2.3–2.9bn of bookings a year. What matters most is how much of Prime members’ DoorDash spending moves, as Grubhub’s own share is only about 3.5 to 5 points. Amazon is room to grow bought with no capital; Delivery Hero is room bought with $13.7bn, which is why the third test turns on it, and on where the rest of the cash has gone.
Cash: generated, spent, returned
Generated
Owner cash is the firm’s measure: free cash flow less stock pay. It flatters a business whose insurance reserves are growing, as that cash is owed to claimants. Take that float out, and it reads:
Owner cash per diluted share, as reported and without the insurance reserve inflow; the second is the firm's calculation.
That is the cash half of the case in one picture: from below zero in 2023 to $3.18. As a share of bookings it has risen from 2.25% to 2.89% over a year on the same basis, with 2.73% at the end of 2025 (the firm’s calculation). Reported, it was flat over the same year, 3.74% to 3.70%, after 4.1% at the end of 2025, as the insurance inflow fades. UK VAT sits outside both lines: Uber has paid $1.9bn in deposits since 2023 that held cash down and may come back, about $1.5bn more is due in Q4, and the outcome in court swings value by about +$0.93 to −$0.73 a share (the firm’s calculation).
Spent
Where the cash came from and went, first nine months of 2026, from the 10-Q's cash flow statement; the split of the securities line is the firm's inference.
Uber raised $5.1bn of net new debt and kept $2.5bn of it in cash; the rest went out with everything else, and money being fungible, which outflow it funded cannot be said. Buybacks of $4.3bn only just exceeded the $3.9bn that went on Delivery Hero’s shares and swaps. Not on the chart: ezCater, $2.3bn in cash agreed in October, and the $1.6bn of smaller purchases in the nine months, which the third test also has to judge.
Returned
Owner cash against buybacks less stock pay, by quarter, the same test as owner cash. An approximation, as stock pay is set at grant.
Set against stock pay, Uber was returning most of its cash through 2025, then almost none in the last two quarters: buybacks less stock pay took $127m of $4,186m of owner cash. Diluted shares did not move in Q3 and are still down 3.5% on the year, so the record is good, and the pause fits saving for the deal. Whether it resumes after close is the other half of the third test; the deal itself is in section 6.
What the price says
Value per share from the firm's model, by bookings growth next year and owner cash margin; shaded cells are below $68.78. These are implied values, not a price target.
The grid is the note’s one number, and every cell is the firm’s model on four inputs: bookings growth fading to 5% over ten years, the margin reaching its column in five, an 11% discount rate and 3% terminal growth after year ten, with robotaxi capital in and Delivery Hero out. Read at today’s 20% growth, between the 18% and 22% rows, it gives about $111 a share on the reported 3.7% margin, about $89 on the 2.9% without the float and about $78 on the 2.5% also without the tax shield Uber’s deferred tax assets provide (the firm’s calculation). Those columns matter more than the ones to the right. The reported 3.7% carries about 0.8 of a point of insurance inflow and about 0.5 of tax shield; the grid’s 2.5% column takes the inflow on the nine months to September, annualised, which is why it is not 2.4%. On the reported margin the price sits between 6% and 10% growth. Read at trips’ 15.6%, the volume section 3 measures, the column without the float gives about $76, and on that column the price sits at about 13% growth; on the 2.5% column, about 17% (the firm’s calculation). The firm’s own case is today’s 20% next year, the middle of Uber’s 18% to 22% guide, fading as the grid does: the $89 on the 2.9% column, an implied return of about 13% a year from $68.78 if the model’s cash flows arrive, which is a reading of the grid and not a forecast.
So the shares are not priced for a collapse, as they look on the reported figures, and at today’s bookings growth the model puts them below value even with the float out. Growth and margin both move it: from $89, growth at 14% reads $71 and a margin of 2.5% reads $78, and a compounder growing in the mid-teens while widening the margin to 3.7% reads about $88. The grid leaves out Delivery Hero, whose effect section 6 sizes, and any earnings from robotaxis. Put into the grid on the firm’s file, no synergies take about $6 a share off the cells near the price, and all of them, taxed, add nothing (the firm’s rough sum).
Where the position stands
Bear case
- Trips per user near flat: Uber is growing on users and bigger baskets, and users are slowing.
- The reported margin leans on a fading insurance inflow and a tax shield.
- On the firm’s rough sum Delivery Hero lowers owner cash per share by about
12%at close, and Uber needs well over half its synergies, in markets under attack, to beat the buybacks forgone. - Waymo’s cars are more than half of the driverless fleet, and its exclusivity ends in January 2028. On the firm’s sum a fleet partner cannot pay Uber’s take at
19rides a day, so the owner that can is the one that can leave.
Bull case
- Owner cash without the float has gone from below zero to
$3.18a share in under three years, with margin rising as growth slowed. - At today’s growth the firm’s model puts value above the price even on the margin without the float.
- Uber has a demand network that fleets want: on its own account, cars on Uber are about
30%busier than first-party fleets, and in mature AV markets rideshare grew7points faster than elsewhere. - Grocery, premium and membership extend the runway, and Amazon now sends its Prime members to Uber Eats.
Uber is earning the name of a compounder on cash, not yet on what it does with it
- Still holds: a business generating real cash, at a price the firm’s model puts below value on the growth the firm expects, near the middle of Uber’s guide, fading over ten years.
- Now rests on: the cash being spent well, on the conditions in section 1’s table: Delivery Hero’s synergies arriving, buybacks resuming after close, and Uber keeping its take as others come to own the cars.
- First test: the offer’s result in November, then Q4 2026 in February, for bookings growth against the growth line in section 1’s table, Mobility’s take rate and the end-2026 autonomy promises.
Spider Eyes Research stays long.
The dials, and what the firm watches next
These are the measures later notes will mark; all but the first are the firm’s calculation from the filed figures named.
| Dial | Q3 2025 | Q3 2026 | Built from |
|---|---|---|---|
| Monthly users | 189m | 215m | Monthly active platform consumers, as reported |
| Trips per user, a month | 6.19 | 6.30 | Trips over monthly users, over three |
| Bookings per trip, Mobility and Delivery | $13.79 | $14.29 | Mobility and Delivery bookings over trips |
| Profit per trip | $0.51 | $0.58 | Adjusted EBITDA less stock-based compensation, over trips |
| Kept of a $20 Mobility ride | $4.24 | $4.13 | Mobility revenue after driver and incentive payments, over Mobility bookings |
| Owner cash a share, no float | $2.49 | $3.18 | Free cash flow less stock-based compensation less the change in insurance reserves, over diluted shares; 2025 against the last twelve months |
| Event | When | What it tests |
|---|---|---|
| Delivery Hero: offer result, then regulators | November 2026, then 2027 | Whether, and when, the deal closes |
| Q4 results | February 2027 | Mobility growth, October’s supply gain, the take rate |
| Waymo’s exclusivity ends | January 2028 | What Uber’s network is worth to a fleet that can leave |
Uber's autonomy ledger, selected, for later notes to mark
| Promise | Due | State, 6 November 2026 |
|---|---|---|
| Zoox on Uber, Las Vegas | November 2026 | Open — was summer, then Q3 |
| Motional driverless, Las Vegas | End-2026 | Open |
| At least 10 cities, as many as 15 | End-2026 | Open |
| About 700 driverless vehicles | End-2026 | Open |
| Nearly 200 driverless in the UAE | End-2026 | Open |
| MOIA rides, Los Angeles | Late 2026 | Open |
| Madrid; Zurich; Tokyo pilot | End-2026 | Open |
| Nuro and Lucid, Bay Area | Q1 2027 | Was 2026. Now driverless from day one |
| NVIDIA, Los Angeles and San Francisco | H1 2027 | Open — later remarks say supervised, 2027 |
| Over 2,000 driverless vehicles | End-2027 | Open |
| 10,000+ driverless vehicles | 2028 and beyond | Open — no year-end number |
How the charts are built
Section 3, what bookings growth is made of: the split is in proportion to users, trips per user and bookings per trip in Mobility and Delivery, with Freight’s effect shown apart; good to about half a point, as Uber reports growth in whole points. Section 4, the $20 ride: a way of reading the percentages, not Uber’s average fare, as Mobility trips are not reported, and bookings include taxes and tolls; Q4 2025 is not reported on this basis. Section 5, the fleet: capacity at 19 trips a day, Waymo’s own figure for its fleet, not rides counted; the faint rows mean Uber gives no upper number; the 120,000 has no date. Section 5, the promises: nothing found does not prove nothing happened, and a delay may be a partner’s or a regulator’s; each launch is scored on its earliest firm date; Wayve’s spring 2026 date is one whose attribution conflicts, and on its August date it would be on time. Section 6, market shares: bases and dates differ; Japan, Korea and Saudi Arabia are vendor estimates, the UK delivery occasions; the markets are selective, weak ones as well as strong; rival dots sit at the middle of the Commission’s ranges. Section 6, delivery margins: adjusted EBITDA before stock pay is the only profit figure comparable across these companies; Delivery Hero and its parts are under IFRS, which leaves lease cost out, and the kept markets are put on Uber’s basis (the firm’s calculation); JET and Deliveroo are 2024; bookings is the base because revenue runs from about 10% of bookings at Instacart to about 39% at talabat. Section 7, owner cash a share: per diluted share; the firm’s model divides by the basic count, 2,038m shares in Q3 against 2,050m diluted, which reads about 2c higher. Section 8, the grid: an 11% discount rate and 3% terminal growth after year ten, fleet capital in and Delivery Hero out; bookings growth fades to 5% by year ten, and the margin moves to its column over five years, then holds. Now is Q3 bookings growth in constant currency, including acquisitions; Guide is Uber’s own for Q4, the only forecast the firm uses here; Trips is Q3.
Sources and method
Uber’s Q3 2026 release, prepared remarks, supplemental data and 10-Q; earlier releases and filings since 2019. Comments on the Q3 call and at the Goldman Sachs conference on 10 September are from machine transcripts, unchecked against official ones. Delivery Hero, talabat, DoorDash, Pony.ai, WeRide and May Mobility filings and releases; California CPUC, New York TLC and Chicago data; European Commission decisions; Consumer Edge and other vendor figures as named. The robotaxi sum’s lines are sourced where a source exists, in Waymo’s, WeRide’s, Pony.ai’s and Lucid’s own figures, Uber’s filings, California’s regulator, Swiss Re’s study with Waymo and the EIA, EPA and AAA, and built up where none does: the driving-system fee, the depot and the robotaxi’s insurance. Rough sums are labelled where they appear.
What this note is. It sets out a valuation of the company and the position the firm holds. Every figure in it is the arithmetic consequence of the assumptions stated beside it — not a forecast, not a target, and not a view on where the price will go — and nothing in it recommends a course of action to anyone.
Position disclosure. Spider Eyes Research and/or its principals hold a long position in Uber Technologies, Inc. (UBER).
Important notice. This document is provided for information and discussion only. It is not investment advice, not a personal recommendation, and not an offer, invitation or inducement to engage in any investment activity. The firm is not authorised or regulated by the Financial Conduct Authority or any other regulator, and holds no licence to provide investment advice.
The firm issues no ratings, recommendations or price targets, and this note does not tell anyone to buy or sell anything. It is written once and published unchanged to everyone who receives it, takes no account of any reader’s circumstances, objectives or tolerance for risk, and creates no advisory relationship with any reader or subscriber.
Figures and estimates in this document are illustrative consequences of the assumptions stated alongside them. They are not forecasts, targets or predictions. Discounted cash flow outputs are highly sensitive to assumptions, and small changes produce large differences in result. Nothing here should be relied upon for any investment decision.
Information is drawn from public filings believed accurate at the date of writing. No representation is made as to accuracy or completeness, and no obligation is accepted to update it. The value of investments can fall as well as rise and you may get back less than you invest. Past performance is not a guide to future performance. Anyone considering an investment should carry out their own research and seek advice from a suitably qualified and regulated adviser.