Uber Technologies, Inc.: pricing 6% growth into a 20%+ compounder

The argument in brief

Uber operates two large marketplaces — ride-hailing and food delivery — plus a smaller freight business. It owns almost no vehicles and employs almost no drivers. Its economics come from matching supply it does not fund with demand it does own, which is why it converts revenue into cash unusually well.

The shares trade near $70 against broker price targets clustered around $100. That gap is not a disagreement about whether the business works. It is a disagreement about two things that have attached themselves to the equity: the unwinding of the Waymo autonomous vehicle partnership, and the €13.0bn acquisition of Delivery Hero, the Berlin-listed food delivery group.

This note argues the market is misreading the first and is partly right about the second.

Five findings

The reported earnings figure is unusable. Uber holds minority stakes in other listed companies — Grab, Didi, Aurora, Delivery Hero — and accounting rules push changes in their market value straight through the profit line. In Q1 2026 that produced a $1.5bn charge and earnings per share of $0.13. In Q2 it produced a $1.6bn gain and EPS of $1.17. The underlying business grew steadily throughout. Uber has to be judged on cash generation, not on headline profit.

On cash, the business is performing strongly. Free cash flow — the cash left after running and investing in the business — passed $10bn on a trailing twelve-month basis for the first time. Capital spending in Q2 was $70m against $14.2bn of revenue, under half a percent.

The apparent collapse in Uber’s revenue margin is mostly an accounting artefact. The take rate — revenue as a percentage of the total value of transactions on the platform — fell 260 basis points year on year. Adjusting for a UK regulatory change that alters how revenue is recorded without changing economics, the fall is 86 basis points. Meanwhile profit per dollar of bookings rose in every segment.

The equity losses that damaged reported earnings were not the autonomy portfolio. They were Grab and Didi, legacy Asian ride-hailing holdings. In Q2 the autonomy holding Aurora contributed a $899m gain. The market has been conflating the two.

The Delivery Hero deal depends entirely on synergy delivery. Uber’s stated purchase multiple of roughly 8x forward earnings includes the full $1.2bn of promised cost savings. Excluding them, the same price is about 20x. Section 5 models what that means across five years.

Financial baseline

Q2 2026 results

Bookings beat guidance of $56.25–57.75bn. Non-GAAP EPS landed a cent below the top of its $0.78–0.82 range. Third-quarter guidance of $58.25–60.25bn implies growth of 18–22% excluding currency, down from 22% delivered — the deceleration behind the market’s cool reception.

Why reported profit misleads, and what to use instead

Two adjustments matter before any of the numbers above can be read sensibly.

The first is equity revaluation. Uber’s minority stakes in listed companies are marked to market each quarter with the change running through profit. This is required accounting, not discretion, but it means reported earnings swing violently for reasons unconnected to operations.

Reported EPS moved 800% across two quarters in which the business grew steadily. The non-GAAP series moved from $0.72 to $0.81. The second describes the company.

The second adjustment is tax. In 2024 and 2025 Uber’s reported net income was inflated by one-off deferred tax credits of $5.8bn and $4.3bn — accounting recognitions of past losses, not cash. Those are now exhausted. The effective tax rate moved from 9% to 26% in a year, and deferred tax assets fell from $10,951m to $10,162m. Comparisons with 2024–25 reported profit are close to worthless.

Cash generation

Two quality notes. Stock-based compensation of $556m is 3.9% of revenue and rising faster than revenue; unvested awards grew from 57.7m to 71.8m shares in six months, so this pressure continues.

More significantly, Uber self-insures much of its US accident liability. It books the cost when a trip happens but pays claims years later, and the gap generates cash. That contribution halved, from $1,487m in the first half of 2025 to $830m in 2026, despite bookings growing 24%. Total reserves stand at $13.3bn — larger than total debt. Either accident costs per trip are falling sharply, reserve assumptions have changed, or payouts have accelerated. These have very different implications, and the quarterly filing does not contain the loss-development table needed to distinguish them. This remains genuinely unresolved until the annual report.

The take rate: decomposing a 260 basis point fall

Revenue as a percentage of bookings fell from 27.06% to 24.46%. Most of this is one accounting change. From 2 January 2026, a UK legislative change meant Uber ceased being the contracting party for private hire journeys in certain UK markets. Payments to drivers moved from being a cost to being a deduction from revenue. Bookings, profit and cash are unaffected; revenue falls by $1.1bn.

The test that matters

If the compression were competitive, profit per dollar of bookings would fall. It rose.

Falling take rate alongside rising unit profit is a different phenomenon from margin erosion, and the market appears to be reading the first as though it were the second.

The Delivery Hero transaction

Terms

A frequent error worth avoiding: the $1.2bn synergy figure is sometimes conflated with the $1.6bn of divestment proceeds. They are separate. Confusing them overstates the synergy case by a third.

On the premium: 127% is measured from a distressed base. Delivery Hero began 2025 near €27, fell through the year, and was valued at roughly €5.8bn in February 2026 against more than €900m of adjusted earnings — about 6.5x for a business growing earnings at 30%. The offer price sits above the 2025 high of roughly €30 and near the mid-2024 peak around €40. Premium to a depressed price is the wrong lens; the absolute multiple paid is the right one.

What Uber is buying

The business is improving under its own management: take rate expanded from roughly 27% to 30%, quick commerce grew over 30% to more than €7.5bn, and its vertical grocery operations reached break-even. Against that, 2026 earnings growth is guided at just 1–6%, cash flow guidance sits below what was achieved in 2025, and the company remains loss-making on a statutory basis.

The two baskets — the strongest argument for the deal

Delivery Hero is being split in two. Uber takes 50 markets; a private equity buyer, SSW Partners, takes 14 as a competition remedy. The two halves are valued very differently.

A 2.4x valuation gap on the same company’s assets, and the composition explains it. Divested are foodora in Austria, Czechia, Norway and Sweden; efood in Greece; Foody in Cyprus; Glovo in Moldova, Poland, Portugal, Romania and Spain; PedidosYa in Chile and Ecuador; and Yemeksepeti in Türkiye. Spain is where courier employment reclassification destroyed the economics. The Nordics are high-wage markets with small populations. Türkiye carries hyperinflation accounting that has distorted results for years. The rest are European markets carrying live employment-law exposure.

Retained are Baedal Minjok in South Korea, talabat across eight Middle Eastern markets, Hungerstation in Saudi Arabia, foodpanda across nine Asian markets, PedidosYa through thirteen Latin American markets, and Glovo across Africa, the Balkans, Central Asia, Ukraine and Italy. Korea and the Gulf are the group’s highest-quality assets: large order values, favourable labour economics, no European employment overhang.

The competition remedy and the quality cull point in the same direction, which is materially fortunate for Uber. If the divested basket runs near break-even, the retained business operates at roughly 2.1–2.3% margin on bookings against 1.8% for the group as reported. Uber is buying Delivery Hero minus its structurally impaired markets, after two years of restructuring carried out at someone else’s expense.

Funding and cash runway

The central question before Q2 was whether Uber would keep buying back shares at $3bn a quarter while committing to a €14.2bn acquisition loan. It answered decisively.

Roughly $5.4bn is already committed to Delivery Hero — $3,502m of shares plus $1,504m of derivatives — before the public offer has opened. Cash fell $2,235m despite $5,078m of first-half free cash flow, and the investing lines show exactly where it went.

This reads as a disciplined programme rather than drift. Uber is buying the asset progressively, funding it from operating cash flow plus a modest increase in borrowing, and using the buyback as the shock absorber. That is the behaviour of a management team intending to protect its credit rating through the transaction.

Runway to completion

Roughly five quarters remain. Liquidity at 30 June was $5,391m, against perhaps $11–12bn of free cash flow over the period.

At the Q2 pace Uber arrives at completion with substantial liquidity and needs perhaps half the committed facility. The principle is simple: every dollar returned to shareholders between now and closing becomes a dollar of permanent debt instead. Management has evidently reached the same conclusion.

Residual risks in the structure

The loan is denominated in euros while Uber’s cash generation is largely in dollars — roughly $1.3bn of exposure across a plausible currency range, with no disclosed hedge. It ranks equally with Uber’s existing unsecured bonds, so current bondholders are structurally diluted. It carries a covenant requiring operating profit to cover interest at least three times; against Q2 operating income of $1,890m and interest of $127m that is comfortable today, but it is the covenant rather than management’s stated intentions that binds if profitability deteriorates.

Two smaller items worth carrying. The $2.0bn term loan drawn in June matures in December 2026 — short-dated, which suggests a placeholder rather than permanent financing. And Uber has paid HMRC approximately £1.4bn ($1.8bn) in disputed VAT assessments, carried as an asset on the expectation of winning on appeal, with further assessments expected. If the appeal fails, that becomes a cost.

Synergies, the five-year bridge and valuation

What the combination creates

The combined group would span 99 markets with pro-forma 2025 bookings of roughly $236bn. The strategic case rests on four claims, in descending order of confidence: geographic fit, since Korea, the Gulf and Latin America are markets where Uber Eats is weak or absent while Uber already runs ride-hailing in many of them; cross-selling, with customers using more than one Uber product generating roughly three times the bookings and profit of single-product users; quick commerce as a distinct product line where Uber has nothing at comparable scale; and industry structure, as consolidation to three global platforms reduces discounting across the sector.

The synergy dependency

Uber states the price implies roughly 8x forward earnings — but explicitly including the $1.2bn of synergies. Working backwards from $13.7bn of equity value plus $4.2bn of net debt, less $1.6bn from the divestment, gives an enterprise value near $16.3bn. At 8x, forward earnings including synergies are about $2.0bn, implying the underlying business contributes roughly $0.8bn.

The headline multiple is manufactured entirely by the synergies. That is not unusual for a large acquisition, but it means synergies are not upside to this thesis — they are the thesis.

The integration dead zone

Timing creates a structural gap. Interest of roughly $820m a year begins at completion in late 2027. The $1.2bn of savings arrives 18 months later, in early-to-mid 2029. Uber carries close to eighteen months of full financing cost against partial delivery. There is also an offsetting commitment rarely netted off: €2bn of investment in Germany over five years, roughly $460m annually, reducing net synergies in the early years to perhaps $700–750m.

Note too that Uber’s claim of being “earnings accretive on close” is a non-GAAP claim — it excludes amortisation of acquired intangibles and deal costs. Statutory results will look considerably worse through 2028.

Five-year contribution bridge

The table models Delivery Hero’s contribution to group results. It is illustrative arithmetic on stated assumptions, built to expose sensitivity rather than to produce a single answer.

Assumptions: retained-book earnings of ~$0.9bn in the first full year growing 8% annually; synergies phased at 40% in 2028 and full from 2029; Germany commitment spread evenly; interest at 5% on €14.2bn equivalent, declining as debt amortises; intangible amortisation of $300m annually.

The conclusion is worth stating plainly. On these assumptions the transaction reduces statutory earnings through 2028 and is roughly neutral on an adjusted basis that year. It becomes clearly positive from 2029 — but only if synergies exceed about half the stated target. At zero delivery it subtracts from earnings indefinitely, because the financing cost alone roughly equals the acquired business’s earnings. Uber’s guidance of high-single-digit earnings accretion by year three is consistent with close to full delivery and nothing less.

Valuation

The base case assumes 2026 free cash flow of roughly $10.8bn, consistent with $5,078m in the first half, growing 15% in 2027 and tapering to 8% by 2031. Delivery Hero is excluded and handled separately above.

Where the $100 consensus comes from

An 11% discount rate with 2.5–3.0% terminal growth produces $93–97. The consensus near $100 is therefore not an aggressive target — it is the ordinary output of a conventional model on unremarkable assumptions, implying slightly faster growth or a discount rate nearer 10.5%. A multiple cross-check reaches the same place: 15x estimated 2027 free cash flow implies about $99.

What the current price implies

At $70 the market prices roughly 6% growth in perpetuity for a business whose free cash flow grew 41% in 2025 and whose profit per dollar of bookings rose 41 basis points this quarter. That gap is the investment debate expressed as a single number.

A caution on method: discounted cash flow is highly sensitive to the discount rate and terminal assumption, and the $82–111 range across plausible inputs is wide. Its value here is diagnostic — establishing what the market must believe — rather than predictive.

Autonomy

The Waymo departure, correctly sized

Waymo, Alphabet’s self-driving unit, is ending its partnership with Uber. The direct financial impact is negligible: the Phoenix deployment was, in Uber’s own description, intentionally limited — around a dozen vehicles. Waymo intends to enter Austin and Atlanta independently from January 2028, with the current agreement running to May 2028. There is no material revenue at risk for roughly two years.

The market is not pricing lost revenue. It is pricing precedent — that self-driving developers use Uber to build demand and then leave once they have local density.

Uber’s structural answer

The relevant distinction is not technological but commercial: whether a partner owns a customer relationship. Waymo, Zoox, Baidu, Pony.ai and WeRide all have their own apps and can leave. Software developers such as Wayve, Nuro, Waabi and Momenta cannot — they build drivers, not demand. Uber has moved down the stack accordingly, and its Rivian agreement makes that fleet exclusive to Uber.

Uber’s autonomy strategy separates into three layers, and it occupies a deliberately different position in each.

The consequence is that Uber captures the customer while holding only minority, unconsolidated interests in the capital-intensive layers. The vehicles sit on other balance sheets, financed by Mubadala, BlackRock, Franklin Templeton and others. Uber’s sub-0.5% capital intensity is preserved by design.

The fragmentation looks deliberate. Different fleet operators in different geographies prevents any single partner accumulating leverage over Uber — reproducing at fleet level the structure Uber already enjoys with individual drivers: many small suppliers, none with pricing power, none owning the customer.

Wayve and London

On 5 August 2026, Transport for London granted private hire licences to Wayve’s autonomous vehicles, completing the requirement that operator, driver and vehicle all be licensed by the same authority. Over 100,000 Londoners joined Uber’s interest list in eight weeks.

That licensing requirement is a regulatory moat favouring Uber. Wayve holds no operator licence; Uber does. In this jurisdiction Wayve cannot carry passengers without an operator partner — a firmer lock-in than merely lacking an app. Wayve’s approach is also the most scalable in the portfolio: it works without pre-mapped roads or fixed geographic boundaries, so the cost of the twentieth city approaches that of the second.

What the balance sheet shows

The equity losses widely read as autonomy bets souring were Grab and Didi. In Q2 they reversed, with Aurora contributing a $899m gain. Committed autonomy capital is largely milestone-gated — up to $1.25bn to Rivian through 2031, $500m to Lucid, roughly $500m to Nuro, $300m to Wayve contingent on London deployment. Uber pays as partners deliver, which makes this programme far more deferrable than the Delivery Hero commitment and the natural flex point if capital tightens.

One detail worth noting: approximately 61% of Uber’s Aurora holding is pledged as collateral against convertible notes and cannot be sold, up from 48% at year-end. Part of the portfolio is less liquid than carrying values suggest.

The counterweight, and the unresolved question

Thirty partnerships is not thirty deployments. Cruise no longer exists. Motional has partnered since 2021 and still runs with safety monitors. May Mobility operates in one Texas city. Fully driverless service on the Uber app appears limited to Abu Dhabi and to Austin and Atlanta with the partner that is leaving.

The critical unknown is what Uber earns on an autonomous trip versus a human-driven one. Its ride-hailing take rate is roughly 25–30%, earned by matching supply it neither owns nor funds. In an autonomous world the partner supplies vehicle, software, insurance and capital, and will negotiate accordingly. It is entirely possible for Uber to win the autonomy race and lose the autonomy economics. Until Uber discloses comparative trip economics, this is the largest unquantified risk in the equity, and counting partnerships does not address it.

The bull case

Stated as the sequence of events required, rather than as assertion.

The compounding continues undisturbed. Bookings grow above 20%, segment margins keep expanding as they did in both Q1 and Q2, and free cash flow reaches $12.4bn in 2027 and $14.3bn in 2028 on capital intensity below 0.5% of revenue. On conventional assumptions this alone supports $93–111 per share with no contribution from Delivery Hero or autonomy. The re-rating requires nothing new — only the absence of disruption.

The pre-funding programme protects the balance sheet. Q2 showed management will flex the buyback rather than the credit rating. At the Q2 pace Uber reaches completion with roughly $12bn of liquidity and needs perhaps half the committed facility. Interest rises far less than the headline €14.2bn implies, and the deleveraging narrative that would compress the multiple never takes hold.

The acquired business is better than group figures suggest. The retained 50 markets carry a higher margin structure because the divested 14 contain the structural problems. If Delivery Hero delivers its guidance and the divestment completes at $1.6bn, Uber inherits a business at roughly 13x with a credible path to 8x as margins converge toward its own delivery economics of 3.84% of bookings.

Cross-selling is the real synergy. Where Uber already runs ride-hailing and adds delivery, the multi-product effect applies directly. Extending Uber One membership across Korea, the Gulf and Latin America is a distribution exercise on infrastructure that already exists — the least execution-dependent part of the case.

Consolidation resets sector economics. Three global platforms replace a fragmented field. Discounting falls and take rates rise across every market simultaneously. This is the largest potential value driver in the thesis and appears in no synergy model, because it accrues to industry structure rather than to any integration workstream.

Autonomy proves to be an option rather than a liability. Milestone-gated commitments mean Uber pays only for delivery. If two or three partners succeed, Uber holds equity in the winners and exclusive distribution on some fleets, without the vehicles touching its balance sheet. If none do, the capital was never deployed.

If the first four hold and the fifth partially materialises, the enlarged group generates $16–18bn of free cash flow by 2029 across 99 markets with structurally reduced competition. At 15–18x that is $240–320bn of equity value against $143bn today. The bull case does not require autonomy to work — which remains its most attractive feature.

What could go wrong

Ranked by probability multiplied by severity.

1. Synergies land late or short. The five-year bridge shows the transaction reducing statutory earnings through 2028 and requiring more than half the $1.2bn to be clearly positive by 2029. At zero delivery it subtracts indefinitely. Uber must roughly double the acquired business’s unit profitability across some 50 markets, starting after completion, having been legally barred from preparing the integration beforehand. Its own acquisition history suggests integrations run late.

2. Autonomy inverts — Uber wins the race and loses the economics. This failure mode is dangerous precisely because it is consistent with every announcement going well. Robotaxis appear on the app in forty cities and the take rate structurally halves. No partnership count protects against it, and the London deployment does not yet test it because a paid safety driver remains in the vehicle.

3. Take-rate compression proves structural. Underlying ride-hailing take rate fell roughly 178 basis points this quarter against 113 in Q1. The compression is accelerating and unexplained. Segment profitability is still improving, so this is not yet an earnings problem — but if the trend continues while the offsetting margin gains exhaust, it becomes one.

4. The tax turn compounds. The effective rate moved from 9% to 26% in a year and deferred tax assets fell $789m. Earnings growth of 35% was achieved despite that headwind; sustaining it as the rate converges further toward statutory levels requires correspondingly faster operating growth.

5. Financing and currency. The euro-denominated facility against dollar cash generation carries roughly $1.3bn of exposure. It must be refinanced into a 2027 bond market of unknown receptivity, ranks equally with existing bonds, and carries an interest-cover covenant. Coverage is comfortable today; it is the covenant rather than management’s stated intentions that binds if profitability deteriorates.

6. Regulatory and execution tail risks. Driver employment classification remains contested across multiple jurisdictions, and the divested Delivery Hero markets demonstrate what reclassification does to unit economics. Merger clearance is required across dozens of jurisdictions and remedies can expand. Delivery Hero must retain staff, merchants and couriers through fifteen months of announced-but-incomplete ownership change. And the halving of the insurance cash contribution, unexplained in current disclosure, may signal weaker cash conversion ahead.

Monitoring checklist

A thesis that cannot be falsified is not a thesis. These are the specific numbers that would confirm or break it.

Q2 2026 scorecard — outcome

Q3 2026 scorecard

Longer-horizon tripwires

Appendix

Quarterly trend

Balance sheet movement

Asset growth is almost entirely the Delivery Hero position rather than operating investment. Property and equipment fell over the period.

Other transactions in progress

Beyond Delivery Hero, Uber acquired SpotHero, a digital parking aggregator, for $617m in April; completed the purchase of Getir’s food delivery business for ~$465m on 1 July; obtained control of Careem Technologies on 30 July, which will consolidate from Q3 and is not in any figures here; and agreed to acquire Blacklane, a Berlin chauffeur service, for ~$550m. It already holds 92% of Trendyol GO in Türkiye — which, with Getir, explains why Yemeksepeti is being divested while Uber builds elsewhere in the same market.

Valuation assumptions

Sources

• Uber Technologies Form 10-Q for the quarter ended 30 June 2026, filed 5 August 2026
• Uber Technologies Form 8-K and earnings release, 5 August 2026
• Uber Technologies Form 10-Q for the quarter ended 31 March 2026
• Uber Form 8-K, 16 July 2026 — business combination and bridge credit agreements
• Uber investor presentation on the Delivery Hero acquisition, 16 July 2026
• Delivery Hero SE FY2025 results, FY2026 guidance and transaction disclosures
• Uber and Wayve announcement on Transport for London licensing, 5 August 2026
• UK Automated Vehicles Act 2024 and Automated Passenger Services Regulations 2026