Uber is a genuinely profitable two-sided network with $11.5B owner earnings and a price/IV of 0.27, yet autonomous-vehicle disintermediation puts the 10-year terminal value in the 'Too Hard' bucket for a Buffett-Munger framework — buy small, with humility.

Plain English

Uber runs the app where you tap a button and a car shows up, and the same app where dinner shows up. After ten years of losing money trying to grow, it finally makes real cash — about $11 billion of free cash flow last year. The price of the stock looks cheap compared to what the business should be worth. The catch: self-driving cars from Tesla and Waymo could either help Uber a lot (cheaper rides, more trips) or replace Uber entirely (you'd just use the Tesla or Waymo app instead). Nobody knows which. So the answer is to be careful.

Thesis

Uber Technologies operates the world's largest mobility and on-demand delivery marketplace, plus a freight brokerage. After a decade of losses, the business turned the corner around 2023 and printed roughly $11.5B of TTM owner earnings (TTM period ending 2025-12-31). The compounder hypothesis is straightforward: Uber sits on three flywheels — riders pull drivers, drivers pull riders, and the same fleet now also delivers food and groceries via Eats. Each incremental city makes the next one cheaper to enter; each new use case (Eats, Grocery, Teen accounts, Ads) lifts driver earnings per online hour, which deepens supply, which tightens ETA, which raises rider frequency. ROIIC over five years is 37.25%, the cleanest signal in the file: marginal capital is compounding at venture-like rates because the network is now scale-mature and adding revenue with little incremental capex. FCF conversion of 70% is respectable for a platform still investing in autonomy, advertising, and international delivery. The valuation is the loud part: scorer base IV is $275, current price $75.12, P/IV = 0.27. Reverse DCF implies the market is pricing in -2.3% perpetual growth, which is plainly wrong for a marketplace gaining 15-20%/year in gross bookings. The 10-year ROIC average of -16.6% is a backward-looking artifact of subsidy-era losses; share count up 31% over a decade reflects that same penance. If you accept the scorer's IV as honest, you do not need a heroic outcome — even a $150 outcome inside three years is a triple. The catch, addressed in moat, inversion, and ten-year sections, is that the very network that justifies the IV is the asset most directly threatened by Waymo and Tesla autonomy. Owning Uber is buying a real business at a quarter of conservative IV — but only if the platform layer survives the AV transition.

Moat

Uber's moat is real but narrower than the IV implies. Walking the five moat types:

Network effects (primary). This is the canonical two-sided marketplace: riders go where drivers are, drivers go where riders are, and a thicker pool on both sides shortens ETAs and lowers prices. In a top-10 US metro Uber typically dispatches a car in under 5 minutes; a sub-scale entrant cannot match that without burning capital. Eats layered on top is a third side of the network — couriers can flex between rides and deliveries, smoothing utilization. Damodaran's framework on switching costs [3] notes that products with low individual switching costs can still be defended if the substitute is structurally inferior; in mobility the substitute (Lyft, local apps) usually has 30-60% longer wait times in mid-tier metros. Erosion risk: in any single dense city the network is contestable — DiDi displaced Uber in China, Grab in SE Asia, Bolt in parts of Europe — so 'global network' is really a federation of city-level networks, and each one is winnable with enough capital.

Switching costs (low). Both consumers and drivers multi-home. A driver running both Uber and Lyft is the rule, not the exception, and consumers price-shop on the lock screen. Damodaran's discussion of search engines [3] applies almost directly: 'there is little cost to an end-user from switching from one engine to another.' Uber's defense is habit, default-app status, and Uber One subscription, which is real but thin.

Cost advantages (modest). Scale gives Uber better insurance pricing (a real edge — insurance is now 12-15% of US gross bookings), a lower customer-acquisition cost on the second use case (an Eats user already in the app costs near-zero to onboard), and routing/matching algorithms trained on more trips than any rival. Ad business piggybacks on this — restaurants and CPG brands now pay Uber for ads inside the app, which is high-incremental-margin revenue that no sub-scale rival can replicate.

Intangibles (brand). 'Uber' is a verb in dozens of languages. Damodaran's brand framework [4] credits brand only when it produces excess returns; here it lowers CAC and supports a small price premium versus Lyft, but it is not a Coca-Cola.

Pricing power (limited). Uber is a price-taker against the marginal driver and the marginal consumer. Surge pricing is real, but every 10% of take-rate increase is litigated by drivers, regulators, and consumer-advocacy groups.

The $10B / 5-year stress test. Could a well-funded entrant — say Tesla or Amazon — kill Uber in five years with a $10B war chest? In any one country, plausibly. Globally, no: $10B buys you maybe two top-five metros at scale, and Uber still spins off >$8B FCF/year to defend. But $10B from a well-organized AV operator (Waymo + Alphabet's balance sheet, or Tesla's ramp) is a different question entirely — see inversion. The platform's network effect is mostly liquidity for human drivers; remove the human and the moat compresses to UI/UX and demand aggregation, both of which are commoditizable.

Moat verdict: NARROW.

Management

Dara Khosrowshahi inherited the business from Travis Kalanick in 2017 and has executed the most important pivot in modern marketplace history: from 'growth at all costs and we'll figure out unit economics later' to 'GAAP profitable, FCF positive, returning capital.' Walk the five capital-allocation choices:

Reinvest in the existing business. Most of operating cash flow continues to go here, and ROIIC of 37.25% over five years says it is working. Eats was the bet that mattered — Khosrowshahi pushed the merger of UberEats and Postmates (2020), then leveraged the combined network through COVID. He divested the things he could not win at high ROI: ATG self-driving sold to Aurora (2020) for stock; Air Taxi sold to Joby (2020); ATG had been burning $1B+/year. He sold Uber's stake in Yandex.Taxi (2021) and trimmed Didi for $1.6B (2024). These are real Buffett-style 'admit the mistake and reallocate' moves.

Acquisitions. Postmates ($2.65B), Drizly ($1.1B, since shut down in 2024 — a $1.1B write-off), Cornershop, Foodpanda Taiwan ($950M, ultimately blocked by regulators in 2024). Mixed record. Drizly was a clear failure. The Postmates deal is harder to call: it bought meaningful share but also bought into the most competitive vertical at peak valuation. Net: B- on M&A.

Debt. Net-debt/EBITDA of 0.97x is conservative for a company of this maturity. Uber refinanced opportunistically into low rates and has been paying down notes since profitability inflected. Interest coverage was negative for years and is no longer reportable as a clean number, but cash on balance sheet exceeds high-yield notes outstanding. Investment-grade rating now (Moody's Baa2). This is responsible.

Buybacks. Uber initiated its first $7B authorization in February 2024, raised it, and has been buying steadily. Average repurchase price has been roughly $65-75. Against scorer base IV of $275, that is buying at ~25-27% of IV — a phenomenal P/IV ratio for buybacks if the IV is correct. This single fact does most of the work in justifying a 'B' grade: management is buying their own stock at the same discount you would. Share count is still up 31% over ten years, but the trend has clearly inflected; net dilution turned into net reduction in 2024.

Dividends. None. Correct call — at 37% ROIIC retained capital is far more valuable than distributed capital.

Communication. Khosrowshahi's shareholder letters are direct, segment-level KPIs are disclosed (gross bookings, take rate, MAPCs, trips, adjusted EBITDA bridge). No funny adjustments beyond standard SBC add-back, which Uber transparently flags. SBC is still high (~$1.8B/year against $11.5B owner earnings — meaningful but not abusive for a tech company). CFO Prashanth Mahendra-Rajah arrived 2024; his tenure is short but commentary has been crisp.

The principal-agent issue is SBC dilution and the buyback offsetting it rather than reducing share count meaningfully — bulls should not expect a Buffett-style 30% reduction in shares outstanding. Realistically, management will keep float roughly flat while compounding earnings.

Capital allocator: B.

Industry

Porter's Five Forces on global ride-hail and on-demand delivery:

Threat of new entrants — HIGH. Software is cheap, the matching algorithms are well-understood, and there is no patent moat. Capital is the barrier, not technology. Every major metro has multiple entrants (Lyft in the US, Bolt and FreeNow in Europe, Grab and Gojek in SE Asia, DiDi in China, Yandex in CIS, Careem/Uber in MENA, 99 and inDrive in LatAm). The interesting wrinkle: AV operators (Waymo, Zoox, Tesla Robotaxi) are new entrants with structurally different cost curves who do not need the human-driver liquidity flywheel. Damodaran's note on disruptive innovation [6] applies directly: established firms 'have too much invested in the traditional form' and may struggle to cannibalize.

Bargaining power of suppliers — MEDIUM-HIGH and rising. Drivers are the supply. They are independent contractors today, multi-home across platforms, and have growing political voice. CA AB5, EU Platform Workers Directive (2024 final text), UK Supreme Court Aslam v. Uber (2021), Massachusetts ballot initiatives — every year more jurisdictions push to reclassify drivers as employees. If reclassification spreads, gross margin drops 200-400 bps in affected geographies and operational complexity rises sharply. AV shifts the 'supplier' from millions of drivers to a handful of fleet operators (Waymo, Tesla) who will have enormous leverage if Uber becomes the demand layer.

Bargaining power of buyers — MEDIUM. Riders multi-home easily, especially in price-sensitive segments. Restaurants on Eats face 25-30% commissions and are vocally unhappy; cities have passed commission caps (NYC, San Francisco). Uber One subscription helps lock in higher-frequency customers. Corporate accounts (Uber for Business) are stickier but a small share.

Threat of substitutes — MEDIUM-HIGH. Public transit, personal cars, e-bikes, scooters, and increasingly direct-to-consumer restaurant delivery (chains running their own apps to escape Eats commissions). The big substitute risk is owned autonomous vehicle — a future where households own self-driving cars that do their own errands and rides for them.

Rivalry among existing competitors — HIGH. Take-rates are bounded by competitor pricing. In any city where Lyft, Bolt, or DiDi are funded, Uber must match incentives. Eats is even more competitive — DoorDash leads US share, Just Eat, Delivery Hero, Meituan globally.

Value pool location and trajectory. The value pool today sits with the platform that aggregates demand. The trajectory question is whether it migrates to the AV fleet operator (Waymo, Tesla) over the next decade. If it does, Uber's position is analogous to Expedia in the OTA stack — meaningful but margin-limited. If Uber successfully becomes the multi-fleet aggregator (already inked deals with Waymo in Phoenix, Wayve in UK, Nuro), it preserves its position. This is not knowable today.

Industry Verdict: Average. Profitable now, but structurally competitive with regulatory and technological tail risks that are larger than typical Buffett-quality businesses.

Inversion

Playing short-seller. No hedging.

The single event that kills this. Tesla launches a fully unsupervised Robotaxi service in five US metros in 2026-2027 at $0.50/mile (versus Uber's all-in cost of ~$1.50/mile for a human-driven trip). Tesla owns the cars, owns the energy cost, owns the demand app, has a brand premium with the early-adopter cohort, and does not need Uber. Within 24 months Uber's mobility GMV in those metros falls 40-60%, take-rate compresses as Uber subsidizes drivers to retain them, and the network's two-sided liquidity — the entire moat — unravels metro by metro. Once the AV genie is out, every well-capitalized OEM (Waymo/Alphabet, Cruise/GM if revived, Zoox/Amazon, Chinese players) follows. Uber's role collapses to either (a) a thin aggregator margin akin to Expedia (high-single-digit take rate on commodity inventory) or (b) irrelevance.

Why the moat is narrower than bulls think. Bulls treat 'network effects' as if Uber were Visa. It is not. The network is city-by-city, multi-homing is rampant on both sides, and the supply side (drivers) has been incrementally captured by regulators. Damodaran [3] explicitly warned about this kind of trap: companies with low real switching costs whose growth is extrapolated despite easy substitution. The 'Uber as a verb' brand is real but priced at $200B+ of moat value when it likely deserves $30-50B. Eats is the most competitive vertical in tech — DoorDash beats Uber in US share, Meituan in China, Delivery Hero in Europe — Uber is a number-two or number-three globally and that position is not improving.

Why management is worse than it appears. Khosrowshahi gets credit for inflecting to profitability, but the inflection coincided with (a) interest rates forcing the entire growth-at-all-costs cohort to shape up and (b) Lyft's strategic retreat from incentive wars. Neither was Uber-specific skill. The Drizly $1.1B write-off, the blocked Foodpanda Taiwan deal ($950M wasted in deal costs and management attention), and the Postmates premium argue against capital-allocation excellence. Buybacks at $65-75 are good if IV is $275 — but if AV disrupts mobility, IV could be $50, and management is buying back at a price they'll regret. The bull's favorite stat — ROIIC of 37% over five years — is partly an artifact of inflecting from huge losses; the steady-state ROIIC is materially lower.

What bulls are extrapolating that won't hold. First, that 15-20% gross-bookings growth continues for a decade. Mobility is tied to urban driving demand which is not a 15% category — eventually growth normalizes to GDP+ in mature markets. Second, that take-rate expands. Take-rate compression is the historically dominant pattern in two-sided platforms once regulators wake up — see Apple/Google app stores, see card networks. Third, that Uber successfully becomes the AV aggregator. Why would Waymo, with a vastly superior unit economics and a working app, choose to give Uber 20% of its margin? Waymo One in Phoenix is already direct-to-consumer. The Uber/Waymo Phoenix partnership is a small early test, not a strategic commitment. Fourth, that owner earnings of $11.5B are repeatable — a meaningful chunk reflects working-capital benefits, equity-method gains on Didi/Aurora positions, and tax assets unlikely to recur at the same magnitude.

Valuation trap (multiple compression / regime change). The bull frame is 'P/IV of 0.27, you can't lose.' But IV at $275 assumes 14% earnings CAGR for a decade. Drop CAGR to 7% (still optimistic if AV disrupts mobility) and IV collapses toward $120-140. Drop CAGR to 0% under a serious disruption case and IV is $50-60. The market may be pricing in -2.3% reverse-DCF growth not because it is wrong but because it is correctly skeptical that the next decade looks like a 14% CAGR. Multiple compression from 16x to 10x P/E on lower earnings is a 50% drawdown without anything dramatic happening.

If I am right, the stock could be worth $35-45 within 3-5 years. The path: 2026-2027 Tesla/Waymo expansion erodes mobility GMV in top metros, take-rate compresses 200 bps, EU Platform Workers reclassification shaves another 150 bps, ROIIC normalizes to 12-15%, and the market re-rates to 12x lower earnings.

Lollapalooza Bias Check

Self-audit on biases active in this analyst right now:

Anchoring (strong). The scorecard hands me a base IV of $275 and a price/IV of 0.27. Once those numbers are on the page, every subsequent thought is anchored to 'this stock is 73% undervalued.' I should remember the IV calculation is mechanical — it extrapolates current owner earnings at a clamped 14% CAGR, which itself was clamped down from 660%. The clamp is doing more work than the model. If I unclamp and use a 6-8% terminal CAGR, IV is closer to $100-130, P/IV is 0.6-0.75, and the margin of safety becomes ordinary rather than spectacular.

Recency bias (strong). Uber's story for the past 24 months has been 'profitability inflected, buybacks announced, FCF rising' — exactly the kind of multi-quarter positive trend that human pattern-recognizers extrapolate forward indefinitely. The opposite recency — five years of losses, going-concern doubts in 2018-2019 — is fading from working memory. Buffett-Munger discipline says one strong year does not establish a moat; ten do.

Authority/social proof (medium). Bill Ackman has a meaningful Uber position. Multiple respected value investors have written it up at similar prices. This is comforting and therefore suspect. Munger: 'When you find that respected people you admire are on the wrong side of a question, look harder.'

Confirmation (medium). I notice myself reaching for stats that support 'undervalued compounder' (37% ROIIC, $11.5B owner earnings) and underweighting stats that don't (10-year ROIC of -16.6%, share count up 31%). The negative metrics are explained away as 'historical artifacts' — which is exactly what the bull narrative requires.

Commitment / consistency (low). I have no prior published view on Uber, so no sunk-cost in the position. This is the bias I am freest from.

Incentive bias (medium). The Buffett-Munger framework rewards finding 'great businesses at fair prices.' Saying 'Too Hard' feels like analytical cowardice. There is professional pressure to have a view. But Munger's own discipline is that 'Too Hard' is a respectable answer and most things should land there.

Deprival super-reaction (low-medium). A 4x potential return is genuinely exciting; missing it would feel painful. Asymmetric loss aversion pulls toward 'just buy a starter position' rather than full discipline.

Lollapalooza synthesis. Anchoring + recency + authority + incentive all push me toward 'Buy.' Confirmation amplifies it. The single bias pushing back is the discipline of the framework itself. Net: my instinct is generous to the long thesis, and I should overweight the AV / regulatory tail-risk arguments to compensate.

10-Year Outlook

Same fundamental business model in 2036? Probably not. Today Uber's economics are dominated by matching human drivers to riders and couriers to orders, with software extracting a 20-30% take. In 2036, in a credible scenario, the dominant ride is autonomous, owned by a fleet operator (Waymo, Tesla, or a Chinese AV champion), and the question is whether Uber is the consumer aggregator or has been disintermediated. Two roughly equally credible futures exist; this is exactly the kind of fork Buffett-Munger discipline says to avoid betting on.

Customer base larger? Almost certainly yes for delivery (Eats, Grocery, Ads continue to grow), uncertain for mobility (autonomous vehicles could collapse per-trip cost and grow trips 2-3x — good for an aggregator, bad for an aggregator that has been disintermediated).

Profit per customer higher? Likely yes if the aggregator role survives — autonomous vehicles remove the largest cost of goods (driver pay), so even at compressed take-rates the gross-profit dollars could be flat-to-up. If disintermediated, profit per customer collapses.

Moat wider? No. The most likely path is moat narrower. The two-sided liquidity advantage versus human-driver competitors fades when AV becomes the dominant supply, and the demand-aggregation moat is narrower than the human-driver-network moat it replaces.

Single biggest threat: Tesla and/or Waymo offering a consumer AV ride app at materially lower price points, building consumer habit directly, and refusing to wholesale through Uber. Secondary threat: regulatory reclassification of drivers as employees in Uber's largest geographies, compressing pre-AV unit economics.

Circle of competence: Uber requires me to predict (a) AV adoption curve and timing, (b) AV fleet operator commercial strategy (direct vs. wholesale), (c) regulatory outcomes across 70+ countries, (d) competitive dynamics in delivery against DoorDash and Meituan. Munger's filter explicitly says 'predicting tech adoption curves / regulatory outcomes' is an auto-fail for circle of competence.

The valuation is genuinely attractive and the business is genuinely better than five years ago. But the 10-year shape is unknowable in a way that puts it outside disciplined competence.

CONFIDENCE: low

Position Guidance

• Recommendation: Too Hard (with optional small position permitted)
• Conviction: low
• Target buy price: $55 (a 27% discount to current, where IV-low of $167 still provides 3x cushion against AV-disruption tail risk)
• Target trim price: $200 (well below scorer base IV of $275; trim aggressively before bull-case fully prices in because the 10-year shape remains unknowable)
• Position sizing: 0-2% of portfolio. The 12-step framework lands at 'Too Hard' on circle-of-competence (auto-fail for predicting AV adoption curves and regulatory outcomes) and on 10-year confidence (LOW). For investors who choose to size in despite the Too Hard signal, treat it as a venture-style bet: small enough that a -70% AV-disruption scenario is survivable, sized to a thesis that can be re-evaluated quarterly as Waymo and Tesla AV economics become legible.
• Trip-wires to revisit: (a) Tesla unsupervised Robotaxi launches commercially in 3+ US metros, (b) Waymo Phoenix partnership terms become public and look unfavorable to Uber, (c) EU Platform Workers Directive forces driver reclassification with material margin impact, (d) take-rate compresses 200+ bps in any quarter.