Uber Technologies

Uber is a matchmaker. On one side are 208 million people a month who want a ride or a meal; on the other, 10.2 million drivers and couriers who want work. Uber's app puts them together in seconds, in more than 15,000 cities, and keeps a slice of every fare. A ten-year-old could explain it: Uber owns no cars and cooks no food; it owns the meeting place, and charges for the meeting. The more riders it has, the faster a driver gets a fare; the faster the fare, the more drivers join; the more drivers, the shorter the wait. That loop is the whole business.

It has become a very good one. In the quarter to June 2026 people booked $58.0bn of trips and orders through Uber, up 22% in constant currency — the fourth quarter in a row above 20%. Over the last twelve months the company generated $10.1bn of free cash flow, having burned $3.4bn in 2020. It is rated investment grade, has bought back $10bn of its own stock since the start of 2025, and in 2026 started counting stock-based pay as a cost in its own headline profit measure.

And yet the shares closed at $67.88 on 1 October 2026 — 32% below their all-time high a year earlier, and barely above the $65.94 low of July. The market is not worried about this year's numbers. It is worried about three things: that robotaxis will make the meeting place unnecessary; that Uber is spending its new cash on a $14.8bn takeover and $10bn of autonomous-vehicle bets instead of on its shares; and that each of the last four quarterly reports was good, but not good enough. This report weighs each of them against the price.

Two lessons for an owner. The first is that this company has already survived one existential threat — the claim, in 2016 and again in 2020, that whoever built the self-driving car would make Uber irrelevant. Uber's answer was to stop building and start aggregating, and in six years the robotaxis have not yet arrived at a scale that shows up in its numbers. The second lesson cuts the other way: the threat has never been closer than it is now, and the first partner to prove it could run without Uber has just announced that it will.

Rides earn the money; delivery earns the habit. Mobility keeps 7.6 cents of every booking dollar as segment profit, double delivery's 3.8 cents — because a ride needs one person and a car, while a delivery needs a restaurant, a courier and a discount. But delivery is what turns an occasional user into a daily one, and the Uber One membership ties the two together: members now account for more than 70% of delivery bookings and about half of all bookings. A member who has paid for free delivery is a member who opens Uber, not a rival, when he needs a ride.

A warning about the revenue line. In the second quarter Mobility revenue grew just 1% while Mobility bookings grew 22%. Nothing broke: from January 2026 Uber stopped acting as the principal in parts of the UK market, for tax reasons, and now records the drivers' share as a reduction of revenue — $1.1bn in the quarter, with "no impact on underlying economics", as the company says. Read Uber through bookings and profit, not through revenue, at least until 2027.

Buffett's 1991 test for an economic franchise is that it sells something "needed or desired", that is "thought by its customers to have no close substitute", and that is "not subject to price regulation." Uber passes the first easily and the third mostly. The second is the whole robotaxi question. Today a rider in Austin has no close substitute for Uber: Waymo's cars come through Uber's app. From January 2028 he will have one, on his phone, run by the best self-driving company in the world. Whether he keeps opening Uber first is not something any analysis can prove in advance.

We score the moat 7. For human-driven rides and for delivery it is among the widest consumer moats we have scored — wider than it was a year ago. The reason it is not 8 or 9 is that the one supplier who could make Uber optional is owned by one of the largest companies in the world and is already showing that it can.

Buffett told a room at Sun Valley in 1999 that the automobile and the aeroplane changed the world and made almost nobody rich: of some 2,000 car makers only three American ones survived, and the money made by all US airlines "since the dawn of aviation" was "zero. Absolutely zero." His lesson was that "the key to investing is not assessing how much an industry is going to affect society, or how much it will grow, but rather determining the competitive advantage of any given company and, above all, the durability of that advantage" (Fortune, 22 November 1999). Charlie Munger made the same point about textile looms: the better machine worked, and "the benefit from the new looms" went "to the people that bought the textiles, not the guy that owned the textile plant."

Applied to Uber, the question is not whether robotaxis will be big. It is who keeps the saving. Our reading: for the rest of this decade the human-driven network keeps compounding, and robotaxis are a rounding error in the volume — Waymo's half a million paid rides a week are about 0.2% of Uber's trips. After 2030 the outcome divides. If robotaxi supply stays fragmented among many makers, Uber is the natural market-maker and could be worth considerably more than today. If it consolidates into two or three giants with their own apps, Uber's take rate shrinks and its moat narrows to the human-driven and delivery businesses. We cannot tell you which. We can tell you that today's price already assumes a good deal of the second.

The pattern. Against companies that do what Uber does, Uber is winning: five times Lyft's size with twice the margin, gaining share in Mobility and roughly holding it against DoorDash. Against companies that could make what Uber does unnecessary, it faces Alphabet, Amazon and Tesla — the first two of them, with a nice irony, among the largest holdings of investors who would never own an airline. According to its 13F filing for 30 June 2026, Berkshire Hathaway holds about $38bn of Alphabet, Waymo's parent — and $5.4bn of Delta Air Lines. It holds no Uber.

Integrity: no veto. The 2026 switch to a profit measure that includes stock-based pay is the opposite of what most technology companies do, and it tells you something about the people. Capital allocation: good, with one large question. Buying back stock at $70–80 has been sensible; spending $14.8bn on Delivery Hero at twice the price paid three months earlier, and pausing buybacks to do it, is a bet that a German food-delivery group with a weaker economic history will be worth more inside Uber than Uber's own shares were. Management says it will add a high-single-digit percentage to earnings per share by year three. We will hold them to it. Owner mentality: the September purchases speak louder than any guidance. We score management and capital 6.

Three subtractions before any multiple. First, GAAP net income: $10.1bn in 2025 included a $4.3bn tax benefit, and the second quarter of 2026 included $1.6bn of gains on Uber's portfolio of stakes in other companies — Didi, Grab, Aurora, Lucid, Delivery Hero — which swing with their share prices. Uber's own non-GAAP earnings, which since January include stock-based pay, are the better measure.

Second, the insurance reserves. Uber sets aside reserves for claims it will pay over years, and while the reserves grow, the cash it has not yet paid out flatters operating cash flow: the build added $830m in the first half of 2026 and $1.49bn a year earlier. Some of that is permanent float, like an insurer's; some will be paid. We take about $1.6bn a year off free cash flow to be safe.

Third, stock-based pay — about $1.9bn over the last twelve months. Uber now deducts it in its own headline profit, and so do we. After all three, the cash an owner can count on is about $6.5–8.2bn a year: $3.20–4.00 a share.

What is the business worth? We discounted owner earnings at 10% over ten years under three stories, and took off about $5 a share of net debt. If Uber remains the aggregator — owner earnings growing ~20% a year for five years, then 10%, and 18 times at the end — it is worth roughly $130–165. In our base case — 15% then 8%, and 15 times — about $85–110. If robotaxis disintermediate it — 8% then 3%, and 10 times — about $45–55. The range is wide because the robotaxi question is real.

At $67.88 the shares sit 20–38% below our base case and 25–50% above the disintermediation case. That is not the 40–50% margin of safety Buffett asks of a business with an unknown this large, which is why we would buy slowly: a first position at today's price, a second below $60. But it is a price at which the market already assumes much of the bad outcome — while the company's chief executive buys $10m of stock at $71 and the business grows bookings at 22%.