Uber Technologies, Inc. (UBER)

Verdict

The verdict, in spine order

Catalyst calendar

Four dates that resolve the debates

Key debates

Five debates, grouped by the spine question

Driver decomposition

Three drivers make the result; the record settles two

• Gross Bookings growth — settled by the record. +18.8% in FY2025, +24% year over year in Q2 2026, a delivered three-year framework behind it. The demand engine is not the open question.
• Take rate — the contested driver. What Uber keeps from each booking (≈25–30% in Mobility, ~19% and rising in Delivery) is where AV supply reprices the model. One point of Mobility take ≈ $809M of after-tax owner earnings — 9.9% of the base (see Reverse DCF). This driver is the page’s turbo trigger and the first price debate above.
• Cost conversion, insurance at the center — settled, trending favorable. Insurance swung from a named headwind (Feb 2024) to a guided tailwind (2026); FCF conversion exceeded 100% of Adjusted EBITDA. The caveat lives in the first debate: part of today’s conversion is reserve accrual, which reverses if severity turns.

The record settles bookings and conversion. The page is about the take rate — and who spends the cash the settled drivers produce.

The kill attempt

Six ways this breaks, stated with the bears’ best numbers

1. The capital-allocation pattern is peaking again — the graded record, not a mood. Drizly was bought for $1.1B at the 2021 top and shut in March 2024. The biggest buyback year (2025, $6.5B) executed at an ~$81.50 average — underwater at $71.27. Now Delivery Hero at €41.50, a ~108% premium to the May 8, 2026 unaffected close, ~$14.8B of equity value — against management’s own standard, in the CEO’s words on Oct 31, 2024: “The best deal is not having to do a deal at all.” In Aug 2026 the same team reported $4B already deployed into Delivery Hero stock in one quarter and buybacks paused, returning in “months, not quarters” (CFO Balaji Krishnamurthy, Aug 5, 2026).
2. Take-rate compression is a priced-in assumption of zero. The flat bar assumes Uber’s take survives intact for a decade. The arithmetic of losing it: one point of Mobility take ≈ $1,079M of revenue ≈ $809M of after-tax owner earnings, 9.9% of the base — the flat bar rises to +8.7%/yr; lose three points and it is +12.1%/yr, requiring roughly +2.8 points of extra bookings growth yearly to offset. Mobility is ~68% of segment operating income, and the $10B AV program meant to defend the take has no filed per-vehicle cost, loss run-rate, or take sensitivity — a numerator with no denominator, for a line still <0.5% of trips.
3. The cash base is partly an actuarial opinion. $13,286M of insurance reserves (≈8.7% of EV) rest on estimates built during fast bookings growth; H1 2026 reserve growth (+6.6%) ran far below bookings growth (+22%), and ~$830M of that accrual sits inside the TTM FCF the bar is struck from. If severity outruns the models, the accrual reverses directly out of FCF — base shrinks, bar rises.
4. The buyback credit rests on a pace that stopped in June. H1 2026 retired 46.6M shares, but ~40% was offset by stock-compensation and employee-plan issuance; the count was flat Mar 2 → Jul 31, 2026 (2,042.9M → 2,042.6M). A program that shrinks on a trailing view while pausing at the moment of maximum spending earns its credit conditionally — and its timing (2025 at ~$81.50, H1 2026 at ~$75.70) shows no price skill.
5. GAAP (generally accepted accounting principles) earnings are a marking exercise, and the finance chair keeps rotating. Q2 2026 net income carries a +$1.6B pre-tax equity-revaluation benefit (+$1.1B Delivery Hero mark, +$899M Aurora, −$437M Didi); FY2024–25 net income was flattered by ~$6.4B and ~$5.0B of tax-valuation-allowance releases. Three CFOs in ~27 months is a continuity debit no single clean print retires.
6. Driver classification is a step-change nobody has priced or quantified. The EU Platform Work Directive, UK assessments and U.S. state turns could re-rate the cost driver in one ruling. No filing quantifies the exposure; a decade of such fights settling at passed-through costs is a history, not a bound.

The bull case

Answering the kill attempt, point by point

1. On Delivery Hero: the acceptance math is already covered, and the promise is testable. Uber’s 24.77% direct stake + ~11.74% via derivatives + Prosus’s irrevocable 16.68% = ~53% economic interest against a 50%-plus-one-share minimum; both boards recommended acceptance (Sep 2, 2026); the €4.5B financing priced at 3.750–5.250%. Conceded: the premium is a fact, and a sweetener above €41.50 hands this point to the bears outright.
2. On the take: fleets need the demand layer more than the demand layer needs any fleet. Partner math (Aug 6, 2024): third-party utilization only needs to beat Uber’s ~20% ex-insurance take — about a 25% uplift — for the marketplace to work for fleets; Waymo’s Austin fleet, exclusive on Uber, ran busier than 99% of human drivers. Partners raise $2.50 of outside capital per $1 of Uber software equity. Conceded: the defense is asserted in units and partnerships, not yet denominated in dollars per vehicle — the Feb 2027 split checkpoint exists because of this.
3. On reserves: intensity is falling while the savings are spent in the open. Reserve growth below bookings growth is what a maturing book looks like; management guides “hundreds of millions” of 2026 U.S. Mobility savings (legislative wins in GA, NV and CA; library record $1M→$60K/$300K) reinvested in fares while U.S. trip growth outpaced. No adverse development disclosed in the window. Conceded: the accrual is inside the FCF base — the bar inherits the actuarial opinion until development proves it.
4. On the buyback: the trailing shrink is real, and the pause was pre-announced, not discovered. Basic weighted shares fell −2.6% year over year in Q2 2026; $15.7B of authorization remained at Jun 30, 2026; management named the pause and its duration on the call itself. Conceded: the timing record is poor — every recent repurchase dollar is underwater at $71.27.
5. On earnings quality: the clean series are the ones this page uses. Operating income (+30% year over year) and FCF (+13%) carry the grade and the bar; GAAP EPS (earnings per share) is flagged, not relied on. The Feb 2026 Form 8-K states the prior CFO’s departure was not a reporting disagreement. Conceded: GAAP net income is unusable for trend until the marks stop dominating it.
6. On classification: a decade of fights has so far ended in pass-through, not prohibition. Insurance repricing moved through the profit and loss statement at ~1-for-1 elasticity (Oct 31, 2024 call). Conceded: genuinely unquantifiable from the filings today — it stays in the gaps, sized by its absence.

The alignment footnote the bears under-use: the CEO bought 141,000 shares at $70.9642 (~$10.0M) on Sep 10, 2026, not under a Rule 10b5-1 plan, two days after a Tesla-robotaxi selloff (Form 4 verified); the COO bought 70,000 shares (~$5.3M) on Sep 4, 2026 — at $71.27 the CEO’s buy is essentially at the money. The trim: officer Jill Hazelbaker sold 28,170 shares (~$2.01M, no plan indicated) on Sep 11, 2026, the day after the buy.

The business

A demand pool with a stakes portfolio stapled to it

Uber runs the world’s largest two-sided mobility and delivery marketplace: 208M MAPC, 3,867 million trips in Q2 2026 (~42M/day), $215.6B of TTM Gross Bookings. Three segments: Mobility (rides; TTM bookings $107.9B), Delivery (Uber Eats; TTM bookings $102.2B, take ~19% and grinding up with advertising), and Freight (TTM bookings $5.5B at a ~100% pass-through take, roughly breakeven — a settled, immaterial drag).

A fourth asset matters to the valuation: a ~$14.0B portfolio of strategic equity stakes (carrying value, plus a $1.5B total-return swap on Delivery Hero) — roughly a tenth of the market cap, stapled to the marketplace. Since Q1 2026 segment profit is Segment Operating Income; Adjusted EBITDA is consolidated only.

Headline quarter

Q2 2026 vs Q2 2025: the quarter that set the guide streak’s terms

Release + Form 10-Q filed Aug 5, 2026. Revenue growth of +12% understates the business: business-model changes (mainly the UK moving from merchant to agent presentation) cut reported growth by ~8 points (−$1.1B to Mobility revenue); underlying growth was ~20%.

The series

Five fiscal years, one inflection, no smoothing

$ in millions. FY = fiscal year ended Dec 31; TTM = twelve months to Jun 30, 2026. Owner earnings = FCF − SBC. WASO = weighted-average shares outstanding.

The rows that decide the thesis (latest periods)

Full series (reference layer)

Growth — Gross Bookings by segment, and revenue

Profitability

Net income is marks-driven — see kill point 5.

Cash

Count

Actual compounding beside the bars: revenue +31.5%/yr FY2021→FY2025, +18.1%/yr FY2023→FY2025; Gross Bookings +20.9%/yr FY2021→FY2025. Owner earnings have no honest five-year rate (negative base) — the series above is the evidence.

Guidance scoreboard

Twelve cells, zero misses — and a framework that already scored

12 of 12 cells inside or above; zero misses in six quarters, re-verified cell-by-cell against the releases. The honest asymmetry: Uber guides one quarter ahead and sets the bar conservatively — the guides given are highly credible, and there is no annual guide to miss.

Long-horizon framework (graded separately): the Feb 14, 2024 Investor Update’s three-year framework — Gross Bookings mid-to-high-teens CAGR (compound annual growth rate), Adjusted EBITDA 35–40% CAGR, FCF conversion >90% — was scored delivered on all three legs at the two-year mark on the Feb 4, 2026 call: bookings ~18% CAGR ($138B → $193B), Adjusted EBITDA ~47% CAGR ($4.05B → $8.73B), 2025 conversion >100% ($9.8B FCF on $8.73B Adjusted EBITDA). One nuance the headline scoring hides: Q1 2025’s EBITDA growth (+35%) was the first quarter to run below the framework’s 35–40% pace — “roughly on pace” before re-accelerating. The final year scores with FY2026 results (Feb 2027; checkpoint 9).

Autonomous vehicles

The language never changed; the capital posture did a U-turn inside it

The transcript series, not the press, documents the pivot:

• May 8, 2024 (origin): AV as a long, hybrid transition; Uber as the demand/matching layer — “going to be very good for the industry. It will be great for Uber” (Khosrowshahi). No capital commitment.
• Aug 6, 2024 (partner math): fleets need only beat the ~20% ex-insurance take via utilization; management expected no substantial AV profits for 5–10 years.
• Feb 5, 2025 (first crack in the asset-light wall): “early on, we will take some balance sheet risk in order to… catalyze the industry… But ultimately, we think all of it is gonna be financialized.”
• Aug 5, 2026 (the pivot, quantified): a $10B multiyear AV investment strategy — equity in software partners (milestone-based; partners raise $2.50 per $1 Uber invests) plus balance sheet for ground infrastructure — with 120,000 vehicle commitments, while AV is <0.5% of trips and Khosrowshahi says “I expect that AV will not be profitable for a few years going forward” (Nov 4, 2025 — note the horizon: “a few years” in Nov 2025 against “5–10 years” in Aug 2024; the expected wait shortened between the two statements).

The direction was never hidden; the check size is new, arriving five months after a Feb 2026 call that quantified no program at all.

Insurance

A six-call swing from headwind to tailwind — with a balance-sheet shadow

Named as a repricing headwind on Feb 7, 2024; easing by Aug 6, 2024; pass-through at ~1-for-1 elasticity Oct 31, 2024; quantified per-trip Feb 5, 2025; flipped to a pricing lever May 7, 2025; and by Nov 4, 2025 → Aug 5, 2026 a guided 2026 tailwind — legislative wins (GA, NV, CA), a driving-insights dashboard, a captive insurer, “hundreds of millions” of expected savings passed to consumers as lower fares.

The shadow is kill point 3: $13,286M of reserves (short-term $3,758M + long-term $9,528M at Jun 30, 2026) are estimates, and the $11,793M of restricted cash collateralizing the program is not netted in EV.

Balance sheet

Sound today, redesigned by choice — and the two commitments, sized

Jun 30, 2026, Q2 2026 Form 10-Q.

Health today: unrestricted cash $4,870M + short-term investments $521M = $5,391M (the only cash netted in EV). Gross debt $12,723M carrying (principal $12,625M; fair value $12.9B) = 1.27× TTM Adjusted EBITDA; net of unrestricted cash, 0.73×.

Every debt tranche (10-Q Note 5, carrying values)

The stakes portfolio: non-current investments $8,759M (Grab $2,020M; Didi $1,900M non-marketable; Aurora $1,763M, ~61% pledged against the exchangeables; other $3,076M) + equity-method $3,773M (Delivery Hero $3,502M carrying vs $3,100M fair value; Careem $147M; other $124M) = $12,532M, plus Delivery Hero total return swaps at $1,504M. Post-period, Uber sold 29.4M Aurora shares for ~$182M (Sep 15, 2026), retaining 157.1M.

Delivery Hero financing, as filed

• Jul 16, 2026 — Business Combination Agreement at €41.50/share cash, ~$14.8B equity value ($13.7B adjusted for prior stake purchases); €14.2B bridge committed; the announcement’s promise, quoted: “gross leverage to remain below 2×,” funded from balance-sheet cash and new debt.
• Aug 6, 2026 — a €4.0B term-loan facility reduced bridge commitments to €10.2B; a new $7.7B revolver (due Aug 2031) replaced the $5.0B facility, undrawn.
• Sep 15, 2026 — €4.5B of euro senior notes issued in five tranches (3.750% 2029 → 5.250% 2046), net proceeds ~€4,453M for general corporate purposes; no filing applies the proceeds against the bridge — as-adjusted net debt ≈ $7,810M, essentially unchanged by the pre-funding.

Close expected H2 2027 (regulatory clearances); management’s timeline: close H2 2027, planning 2028, migrations 2029; claimed synergies >$1.2B (company framing, scored at close).

Concentration, priced against EV ($152,908M)

• Delivery Hero equity value ~$14.8B = 9.7% of EV.
• The AV program $10B = 6.5% of EV = 122% of TTM owner earnings = ~99% of one year’s TTM FCF.
• Summed — same team, same confidence peak, same FCF cushion — $24.8B = 16.2% of EV ≈ 3.0× TTM owner earnings.
• The stakes portfolio ($12,532M + $1,504M swaps = $14,036M) = 9.2% of EV, at carrying values and marks: the one visible mark gap (Delivery Hero equity-method, fair value $402M below carrying) is −0.3% of EV — small today, and why the stakes-adjusted lens in the Reverse DCF excludes the portfolio at carrying value, not at a haircut.
• Insurance reserves of $13,286M = 8.7% of EV sit on the liability side.

Pro-forma leverage, honestly: the full original €14.2B bridge drawn on top of today’s debt would be ~$28.6B ≈ 2.85× TTM Adjusted EBITDA (2.57× including the Delivery Hero perimeter’s CY2025 EBITDA of $1.1B — mixed periods, labeled). The filed stack explains why it won’t happen that way (€18.7B of facilities over-commit by certain-funds design; balance-sheet cash funds part; close is a year away) — but “below 2×” is the number to hold them to at close, and no filing yet proves it.

Reverse DCF

The price asks for duration, not heroics

House stack: 10% required return, 10-year window, 2% terminal growth, terminal value on the filed model’s Gordon form, no exit multiple, solved on EV of $152,908M. Base = owner earnings = FCF − SBC: TTM FCF $10,116M − SBC $1,938M = $8,178M (SBC = 3.5% of TTM revenue, its own line in the series). The model’s basis is unchanged from the prior edition; at the prior price of $69.29 it reproduces the prior bars exactly, so the +0.4-point move in the bar is a pure price effect.

• CAP (second inversion): ~5.0 years at +12%/yr; ~4.0 at +15%; ~3.0 at +20% — after which growth fades to 2% forever.
• Terminal value = 53.2% of EV; implied terminal EV/EBIT ≈ 9.4× — inside the mature 8–14× band; nothing exotic is assumed about the end state.
• Sensitivity (flat bar): hurdle 9% → +5.4%/yr; 10% → +7.3%; 11% → +9.1%; 12% → +10.9% (at 2% terminal growth). Terminal growth 1%/3% at 10% → +8.2% / +6.2%.

Turbo trigger — the take rate decides; the other drivers are dismissed numerically. Losing 1 point of Mobility take lifts the flat bar to +8.7%/yr; losing 3 points, to +12.1%/yr (level-shifted base, filed solver). Bookings growth cannot bind first: the grid below shows even +4%/yr revenue growth justifying the price at a 22% mature margin. A non-take margin collapse would have to reach 8% — barely half today’s 14.8% owner-earnings margin — to break the price at delivered growth rates. Take compression is the one driver both plausible and sufficient.

What you must believe — breakeven grid

Implied $/share; 10-year revenue CAGR × mature owner-earnings margin, margin ramping linearly from today’s 14.8%; filed model’s terminal machinery; ✓ = justifies $71.27.

18 of 25 combinations clear the price; the central cell (12% growth × 15% margin — close to today’s delivered margin at two-thirds of FY2025’s bookings growth) implies $102.92. The price fails only where growth halves and the margin compresses — which is the take-rate scenario, stated in the grid’s own language.

Buyback credit — granted at 2%/yr, conditionally. Demonstrated on the trailing view (basic weighted shares −2.6% year over year in Q2 2026; H1 retired 46.6M shares). But Q2’s pace was $518M with zero June purchases and the count was flat Mar→Jul. Credit the arithmetic; do not credit the run-rate — checkpoint 2 carries the re-basing trigger.

Moat — Fisher’s questions

What is answered, what is thin

1. Runway and the successor product Answered — Bookings +18.8% in FY2025 across 208M consumers; the successor lines (advertising, the AV layer, Uber One bundling) are in build, not in pitch decks.
2. Innovation productivity Thin — Corporate general and administrative costs plus platform research and development rose +18% year over year in Q2 2026 against +12% reported revenue (~+20% underlying); no filing gives an output-per-dollar measure. The spend is asserted productive, not demonstrated.
3. Go-to-market quality Answered — Density is the salesforce: shorter waits win riders, fuller order books win couriers, city by city; Uber One’s 50M members ride and order — a cross-category relationship neither Lyft nor DoorDash replicates.
4. Margin level — and what defends it Answered — Operating margin −22.0% (FY2021) → 12.1% (TTM). The defense on the record: insurance legislation won in three states, pricing as a lever, a captive built. The caveat: part of the margin is reserve accrual (kill point 3).
5. People Answered — A two-sided operations bench running 42M trips a day through a full cost cycle; driver supply’s fragility is regulatory (checklist), not organizational.
6. The industry-peculiar edge Answered — The insurance/regulatory/operations stack — captive, safety and driving data, decade-long regulator relationships — is what a pure AV or delivery rival cannot copy quickly.
7. Horizon and dilution economics Answered — SBC down from 6.7% to 3.5% of revenue; the count now shrinks on the trailing view; the framework was set for three years and scored in the open.
8. Candor Answered — Integrity is graded as the capstone in Management below. The thin item (2) and the reserve caveat (4) route to the Downside Checklist, not to prose.

Management

Grade: B− — the cap is behavior, not honesty

Re-verification (four checks, run Oct 16, 2026): (a) Form 4 sweep, Oct 13–16 — exactly three filings, all Oct 14: routine director restricted stock unit (RSU) grants (Burns 332, Arora 315, Thain 350 units, code A) — no officer activity, no buys or sells; (b) latest proxy unchanged in substance (DEF 14A Mar 23, 2026; the Apr 20, 2026 amendment corrected a typographical error only); (c) no filing since the Q2 10-Q updates the Jul 31, 2026 count or the buyback record — the Nov 4 10-Q remains the test; (d) no calls since the Aug 5, 2026 Q2 call. Verdict: B− stands.

The Fisher skeleton, as the grade’s evidence

1. Guidance behavior Answered — A−: 12/12 cells re-verified; the Feb 2024 framework delivered on all three legs; guides are one quarter ahead and conservative, and the page grades that asymmetry rather than hiding it.
2. Execution Answered — A−: margin −22% → +12.1%, FCF −$743M → $10,116M, the insurance cycle managed end-to-end.
3. Executive coherence Thin — flagged — Three CFOs in ~27 months; the Feb 2026 Form 8-K states the prior departure was not a reporting disagreement.
4. Capital allocation Contested — C+: argued in full at kill point 1 and bull point 1 — the divestiture record (ATG, Jump, Zomato, Yandex, the Sep 2026 Aurora block) sells early and well; the buying record pays peak prices at peak confidence.
5. Alignment Answered — B−: the CEO’s ~$70M outright stake is ~6–7× the proxy’s 10×-salary guideline; hedging/pledging prohibited; the 2025 performance restricted stock unit (PRSU) margin tranche paid 0.00% (the plan bites); say-on-pay ~93.8%; SBC sits inside the incentive metrics — against which stand the Sep 2025 plan sales (450,000 shares, ~$43.7M) and Hazelbaker’s sale.
6. Integrity, the capstone gate Answered — No disclosure failure surfaced in any sweep; the buyback pause was pre-announced with its duration on the call itself.

The cap on this grade is behavior — what they buy — not honesty about what they say.

Pabrai Downside Checklist

The closing veto gate

The verified record, run against the thesis before the verdict hardens.

• Valuation Pass One flag — PASS: the bar (+7.3%) sits far below every delivered rate since profitability. FLAG: the base carries ~$830M of H1 reserve accrual — capitalizing a flattered year is the classic mistake available here if severity turns.
• Leverage Pass today Flag pro-forma — PASS: net leverage 0.73×. FLAG: the fully drawn Delivery Hero stack computes to ~2.85× against a “below 2×” promise — leverage misjudged at the commitment, not the draw, is the mistake this category exists to catch. (This flag is the verdict box’s top unverifiable.)
• Management & ownership Flag — The CEO and COO bought ~$15.3M of stock in Sep 2026 at ~today’s price (pass) — against buyback timing underwater on every recent dollar and three CFOs in ~27 months (flags).
• Moat Flag — AV disintermediation is unquantified in every filing: a structurally cheaper supply entrant re-pricing the incumbent’s take is the Dexter Shoes shape — the loss arrives as a margin, not a headline. (Stated in the verdict box’s moat line.)
• Personal bias Flag — the desk’s own — The “peak-confidence pattern” rests on two or three episodes; pattern-matching can manufacture the narrative it claims to find. The disconfirming evidence — covered acceptance math, termed-out financing, insider buying — sits on this page beside the pattern, deliberately.

What would change the read

Dated checkpoints with thresholds

1. Nov 4, 2026 — Q3 print vs the guide. PASS: Gross Bookings ≥ $58.25B and Adjusted EBITDA ≥ $2.86B (the guide low ends; non-GAAP EPS guide $0.84–0.88). FAIL: either below its low end — the first miss in seven quarters re-tests the A− guidance grade.
2. Nov 4, 2026 — the Q3 10-Q buyback table. PASS: Q3 repurchases ≥ $1,000M (re-acceleration toward the H1 pace) with the count below 2,042.6M on the new cover. FAIL: repurchases < $518M (Q2’s pace) or a count at/above 2,042.6M — a second flat quarter re-bases the credited bar (+5.2%) to the flat bar (+7.3%).
3. Nov 5, 2026, 24:00 Frankfurt — Delivery Hero acceptance (binary). PASS: the 50%-plus-one-share threshold met at €41.50 with no price increase. NEUTRAL-OPEN: an extension without a sweetener. FAIL: any increase above €41.50 — the capital-allocation pattern confirming itself in real time.
4. Delivery Hero financing — by the Q4 print (Feb 2027). PASS: a quantified leverage figure for the close at or below 2×, or further bridge take-outs below €10.2B. FAIL: the bridge still at €10.2B and no quantified restatement by Feb 2027.
5. AV program specificity — by the Q4 2026 call (Feb 2027). PASS: a dollar split of the $10B across its two named buckets (software equity; ground infrastructure). FAIL: still two buckets and no figures.
6. Insurance — Q3 and Q4 2026 prints. PASS: no prior-year adverse development ≥ $250M disclosed in either quarter, and the 2026 savings claim reaffirmed at “hundreds of millions” or better. FAIL: adverse development ≥ $250M in either quarter.
7. The ezCater question — by Nov 4, 2026 (binary). A Form 8-K confirming terms (price + funding source) resolves it; no 8-K and no denial by the print leaves it reported at zero weight. (Single mention: Dow Jones reported a $2.3B deal on Oct 6, 2026; no filing through Oct 16 confirms it.)
8. Freight — Nov 4, 2026. PASS: Q3 bookings growth ≥ +15% year over year and segment operating income ≥ $0. FAIL: growth < +5% or a loss worse than −$50M — Q2’s +25% bounce was a quarter, not a trend.
9. Framework final year — Feb 2027. PASS: three-year bookings CAGR ≥ 15%, Adjusted EBITDA CAGR ≥ 35% (the framework’s own corrected range), FY2026 FCF conversion ≥ 90%. FAIL: any leg below its framework floor.

Update protocol: this page is re-scored on the Nov 4 print and Nov 5 acceptance outcome (checkpoints 1–3, 6, 8 first); the next edition states each threshold’s result before changing any bar.

Corrections in this edition

Dated, stated for the record

Carried from the prior edition — the five corrections to the Oct 15 dive, restated for the record: FY2025 Adjusted EBITDA is $8,730M (not $8,773M), TTM $10,043M (not $10,086M) — a $43M error in the prior dive’s quarterly sum. The pre-bid Delivery Hero stake is 24.77% (not ~24.99%), plus ~11.74% via derivatives. The bridge is €10.2B of commitments, undrawn (not €14.2B), after the Aug 6, 2026 €4.0B term-loan take-out. The ~€4.5B euro bond is filing-verified (priced Sep 9, closed Sep 15, 2026). Prosus’s irrevocable undertaking is Tier 1.

New in this edition (Oct 16, 2026): the Feb 2024 framework’s Adjusted EBITDA leg is corrected to 35–40% CAGR (the prior edition printed “high-30s to ~40%”; the library record is 35–40%) — the delivered ~47% clears either range, so no verdict moves. The CA uninsured-motorist figure is corrected to the library’s three-state record ($1M→$60K/$300K across GA/NV/CA; the prior edition printed “$1M to $300K,” selecting one endpoint). The AV profit-horizon tension is now stated (Aug 2024: no substantial profits for 5–10 years; Nov 2025: “not profitable for a few years” — the horizon shortened between statements). The prior edition’s completeness note counted “four DCF lenses”; five are published.

Gaps

Stated, not smoothed

• The Delivery Hero offer document’s full text (a German takeover-law document) is not on EDGAR and was not retrieved; every figure here is anchored to the Jul 16, 2026 Form 8-K/announcement and the Aug 27, 2026 publication release. Leverage language inside the offer document itself is unverified.
• No post-Jun-30, 2026 buyback or share-count data exists until the Q3 10-Q (Nov 4); the 2026 Term Loan had $1.0B undrawn at Jun 30 with borrowings permitted only through Aug 2, 2026, and no filing confirms or denies a further draw.
• The $10B AV program has no per-partner or per-category dollar breakdown in the call record — two buckets only; the 120,000 vehicle commitments are not broken out by partner.
• TTM weighted-average share counts are derived (day-weighted from quarterly 10-Q figures), not filed as single TTM figures; fiscal-year counts are as filed.
• Waymo partnership-status changes, AV operating scale, and “surveillance pricing” scrutiny are reported — color, never model inputs. Delivery Hero trading ~11–12% below the offer in October is likewise reported market color, not acceptance evidence.

Sources

Primary sources, ordered by exhibit

• Figure 1 — Gross Bookings, annual (Qualtrim export). Figure 2 — Free Cash Flow, annual (Qualtrim export). Figure 3 — Revenue, annual (Qualtrim export).
• SEC EDGAR (CIK 1543151): FY2021–FY2025 10-Ks (FY2025 accn. 0001543151-26-000015) · Q1/Q2 2026 10-Qs (Q2 accn. 0001543151-26-000032) · earnings releases on Form 8-K, Q4 2024–Q2 2026 · DEF 14A (Mar 23, 2026, accn. 0001308179-26-000125) + DEFR14A/DEFA14A (Apr 20, 2026) · Form 4s (Khosrowshahi purchase Sep 10, 2026, accn. 0001184237-26-000008; Macdonald Sep 4; Hazelbaker Sep 11; director RSU grants Oct 14, 2026) · Delivery Hero 8-K (Jul 16, 2026, accn. 0001552781-26-000382) · term-loan/revolver 8-K (Aug 6, 2026) · euro notes FWP/424B2/closing 8-K (Sep 2026, accn. 0001552781-26-000486) · Aurora 13D/A (Sep 17, 2026) · AGM results 8-K (May 8, 2026).
• Company documents: Investor Update framework (Feb 14, 2024); Delivery Hero offer publication release (Aug 27, 2026); boards’ reasoned statement (Sep 2, 2026).
• Transcripts: the desk library’s 12 full-call notes, Q3 2023–Q2 2026.
• Price: Yahoo Finance chart API, re-verified Oct 16, 2026.

Full accession-level source list: dossier evidence files.