Primary sources used in this piece include Uber’s 2019 Form S-1 (SEC EDGAR); Garrett Camp’s own Medium post “The Beginning of Uber” (2017), which republished the original late-2008 UberCab pitch deck; Bill Gurley’s Above the Crowd essays; Susan Fowler’s February 19, 2017 blog post; contemporaneous journalism from TechCrunch, Forbes, TIME, and Fortune; first-person accounts from early Uber employees and from Sidecar founder Sunil Paul; and Brad Stone’s “The Upstarts” (2017). Where something is verified by a filing or a participant’s own published account, I quote it. Where something is single-sourced or disputed, I say so in the text and the notes.
Prologue: The Orders on the Blog
In the third week of October 2010, a startup barely three months old as a Delaware corporation received cease-and-desist orders from two regulators at once: the taxi division of the San Francisco Municipal Transportation Agency and the California Public Utilities Commission. The orders, reportedly dated October 20, threatened fines of up to $5,000 per violation and potential jail time. The company, UberCab, ran a small app that summoned a licensed black car in San Francisco with your phone.
Most companies comply with letters like that, or quietly lawyer up. UberCab did neither. It published both orders in full on its own blog, dropped the word “Cab” from its name — a rename made, TechCrunch reported, “to avoid the appearance of marketing itself as a taxi business”; the Delaware entity, incorporated as Ubercab, Inc. in July 2010, formally became Uber Technologies, Inc. the following February — and kept operating. The coverage, in TechCrunch’s words, “sparked an outcry from anti-regulation camps and fans of Ubercab web-wide.” The regulators had handed the company its first press cycle; the company had discovered that enforcement could be converted into marketing.
That same month, it closed its seed round.
Almost everything Uber would become is visible in that one October week. Treat legality as a negotiation. Treat confrontation as distribution. Don’t stop.
Part I: The World They Were Born Into
Uber’s founding window, 2008 to 2010, stacked three environmental shifts; the company is not explicable without all three.
The first was the phone. Apple opened the App Store in mid-2008, and GPS chips were becoming standard in smartphones at almost the same moment. For the first time, an ordinary consumer carried a device that knew where it was and could broadcast that location to software in real time — and car dispatch is, at its core, a matching problem between two moving points on a map. The idea for UberCab dates to August 2008, within weeks of the App Store opening; the product was impossible eighteen months earlier.
The second was the financial crisis. The recession left professional drivers — limo operators, town-car fleets, later anyone with a decent car — hungry for income, black cars idle between corporate bookings. The earliest pitch that circulated among angels described exactly this arbitrage: “a service that turns the downtime of town car/limo services into cab service,” as one prospective investor recorded it. Cheap, motivated supply was a macro condition, not a company invention.
The third was money. The policy response held interest rates near zero for most of a decade, and capital that couldn’t find yield went hunting for growth. Uber became the single largest beneficiary of that regime — more than $5 billion raised by 2015, by contemporaneous accounts.
Beneath all three sat the target: taxis. Municipally regulated, supply-capped by medallion systems, fragmented into thousands of small fleets running dispatch technology decades old. The incumbent was not merely weak — its entire defense rested on the assumption that enforcement agencies had the staff and will to act.
Part II: The Founders
Garrett Camp, the originator
The story most people know is that Uber was born on a snowy night in Paris in December 2008, when Travis Kalanick and Garrett Camp couldn’t hail a cab. Uber’s own corporate history tells it that way. The founder’s account does not.
In August 2017, Camp published a short Medium post marking, in his words, “the ninth anniversary of the idea of Uber (originally UberCab, in Aug 2008)” — and attached “the very first pitch deck we created in late 2008.” By the founder’s own published account, the idea predates the Paris trip by roughly four months. Stone’s reporting fills in the sequence — Camp conceived the idea in San Francisco, worn down by the city’s scarce cabs, reportedly registered the ubercab.com domain in early August 2008, and formed a UberCab LLC that November. The December 2008 LeWeb conference in Paris is where Kalanick got drawn in, not where the idea was born. Camp’s post never mentions Paris at all.
Camp could afford to chase the idea: he had co-founded StumbleUpon and sold it to eBay in 2007. His own telling includes splitting an $800 private-driver bill one New Year’s Eve — an anecdote that defines the original customer: someone for whom $800 was an annoyance.
Travis Kalanick, the operator
Kalanick’s pre-Uber record is instructive. His first company, Scour, a late-1990s peer-to-peer search and file-sharing venture, was sued by the media industry and ended in bankruptcy. His second, Red Swoosh, founded in 2001, he ground through six lean years before selling to Akamai in 2007 for a reported $19 million. Neither outcome made him famous. Both made him something more dangerous: a founder who had been sued by giants, nearly broke for years, and had concluded that survival belongs to whoever refuses to flinch.
He joined UberCab as something closer to an advisor — Stone has Camp pushing the idea for months against Kalanick’s initial reluctance. The personality that would define the company is visible in documented moments across this story: dictating the terms of his own financing rounds, marching into a regulator’s office to demand a competitor be shut down, berating one of his own drivers on camera. All of it is on the record, and for years investors treated it as a feature.
The service was initially run by neither founder: Ryan Graves became UberCab’s first employee and first CEO by replying to a Kalanick tweet — the future giant hired its first chief executive over Twitter. Kalanick took the seat before the end of 2010 and held it until June 2017.
Part III: The Thesis and the Pitch
The original UberCab deck — the late-2008, roughly 25-slide document Camp published in 2017 — lets you check the founding thesis against the audited outcome, no narration in between.
The thesis was modest and precise: a premium, cashless, one-tap black-car service — faster than a taxi, cheaper than a town car, dispatched by software. The market slide, per Forbes’ slide-by-slide review, put the “initial market assessment” at a $4.2 billion opportunity in U.S. taxi and limousine services. The best-case scenario projected $1 billion in annual revenue. The self-described more realistic scenario: 5% of top U.S. cities and $20 million in profits. The founders claimed the service would be “profitable by design.”
By the 2019 IPO, Uber’s revenue was $11 billion — eleven times the deck’s best case — with operations in 100% of top U.S. cities and roughly 30% of the world’s countries. The founding document undersized the company’s own market by roughly two orders of magnitude. As for “profitable by design”: in the year before the IPO, the audited operating loss was $3.0 billion.
The seed: an uncontested deal
Rob Hayes of First Round Capital found the company in mid-2010 because he followed Camp on Twitter — First Round had backed StumbleUpon — and noticed him tweeting about something called UberCab. Hayes reached out himself with an email of just a few words; CNN’s account records it as “I’ll bite,” sent June 15, 2010. In October 2010 — the same month as the cease-and-desist orders — UberCab closed a $1.25 million seed led by First Round, with Chris Sacca’s Lowercase Capital, Founder Collective, and several angels, at a reported $4 million valuation.
Hayes was honest about what he thought he was buying: “I saw it as bigger than the black car market,” he said later, “but I absolutely did not see it as the amazingly global and massive company that it is today.” By his own account “it wasn’t a terribly competitive round.” Angel Mike Walsh put in $10,000, reasoning the worst case was that “somebody would buy this thing.”
The passes are as well documented as the checks. John Greathouse, offered the deal in late 2010, later published his reasoning: the pitch was a black-car utilization play, and the sophisticated objection of the day was “I don’t see it expanding beyond a few small cities with strong tech communities.” His partner flagged the fresh cease-and-desist as a deal risk; Mark Cuban reportedly passed on valuation and regulatory grounds. Greathouse also recorded the era’s bear case — “If they try to move into Philly, New York or Boston, they’re going to get their throats slashed” — exactly right about the fight, exactly wrong about who would bleed.
In February 2011, Benchmark led an $11 million Series A at a reported valuation around $60 million, and Bill Gurley joined the board. Gurley’s thesis — written out fully in 2014 — was a three-part liquidity network effect: more riders mean “the shorter the pick up times,” denser coverage, and higher driver utilization, enabling lower prices, recruiting more riders. It was the correct model of the business. What even Gurley couldn’t size in 2011 was the market underneath it. Nobody could — and that is the point.
Part IV: The Growth Machine
This is the center of the story, so I’ll take it mechanism by mechanism — from first-person operator accounts, contemporaneous reporting, and the S-1, with flags where memory has inflated.
The first city: premium, legal, word of mouth
UberCab opened in San Francisco in mid-2010 — accounts put the beta around May–June, after a reported early New York test with a handful of cars — and it launched legally. The original service connected members to licensed black-car drivers, which California livery regulations permitted. Early growth was nearly all organic: a striking product experience (press a button, watch the car crawl toward you on the map), early adopters primed to evangelize, prices that filtered for users who valued time over money. By 2012, per statements Kalanick made that year and Gurley later cited, roughly 600 black cars were active in San Francisco and the business was growing about 20% month over month — a rate Stone has Uber still holding at the 2013 round. Every other mechanism in this section exists to feed that number.
The city-launch playbook
Uber’s real operational invention was not the app. It was the repeatable city launch — market entry converted from an art into a checklist.
By ex-employee accounts, the playbook was a living document of hundreds of tasks, revised after every launch and executed by small traveling “launcher” teams — a model associated with early employee Austin Geidt. The sequence was constant: drivers, then riders, then liquidity.
On the supply side, launchers acquired drivers with unglamorous tactics — one early employee describes cold-calling limo bases from phone lists — layered with an incentive stack of hourly earnings guarantees, sign-up bonuses, and driver referral rewards. The same account puts the supply threshold to open a city at 20 to 30 drivers, though “the magic number was 50.” Dollar amounts are thin: one early-employee account puts driver referral bonuses at $200–$500 (up to $1,000 in San Francisco); retrospective accounts describe $30–35 hourly guarantees — directionally credible, individually unverified.
On the demand side: give-get referral promo codes, free rides for the first two days after a city opened, and “rider zero” events where a local celebrity took the city’s first trip. Demand was spiked around events, weekends, and holidays — the moments taxis failed worst.
The playbook compounded. Paris — the city of the founding myth — became Uber’s fifth city and first international market in 2012, and by mid-decade the machine was opening cities at a pace measured in days. (Insider recollections of the pace run hot; see notes.)
Surge pricing: the machine’s thermostat
Uber’s most infamous mechanism has an origin story from exactly one person, so I attribute it as such. Kevin Novak — by his own account employee #21 and the second data hire, who says he “personally implemented the entire surge pricing product” and ran the surge team until mid-2015 — writes that dynamic pricing was first implemented in late 2011, built for the spikes that hit every Friday and Saturday night and holidays like New Year’s Eve.
The problem it solved was physics, not greed. In Novak’s telling, when demand spiked in one neighborhood, each successive ride was dispatched to a car farther away, until the fleet’s carrying capacity collapsed and nobody could get a ride. Raising prices rationed demand and — more importantly — moved drivers toward the surge zone. Novak’s claim is that early on the fleet completed “two or three times as many trips per hour” with dynamic pricing, and that the revenue impact came mainly from throughput, not the surcharge. Surge was a liquidity engine wearing the costume of a price gouge.
Capital as a growth mechanic
By 2013, fundraising was not a means of funding the playbook; it was a component of it. The financing round that year, chronicled by Stone, shows who held the leverage: Google invested $258 million; TPG put in $88 million by buying shares directly from Camp, with a ratchet protecting it if Uber’s valuation ever fell below $2.75 billion; Jay Z agreed to invest $2 million and wired $5 million. Kalanick, colleagues told Stone, set the terms himself. In June 2014 Uber raised $1.2 billion more at a $17 billion valuation.
That capital bought subsidies — rides priced below cost, guarantees above market earnings, bonuses on both sides — and the subsidies bought the curve. Mechanically, Uber’s growth machine converted zero-interest-rate capital into market share.
The pivot Uber fought before it made it
The most consequential product decision in Uber’s history — UberX, the shift from licensed black cars to ordinary people driving their own cars — was not Uber’s idea. The people whose idea it was wrote it down.
Sidecar launched peer-to-peer ridesharing in San Francisco in February 2012 (live by February 7, per founder Sunil Paul), using a donation-based payment model deliberately designed to thread the regulatory needle. Lyft followed that summer. Uber’s first response was not to copy them — it was to try to have them killed. In August 2012, per the Harvard-published account, Kalanick went to the CPUC in person: “I’m here to ask you to shut down Lyft. They are illegal and you guys have to shut them down.”
The regulators declined. Lyft and Sidecar drew cease-and-desist orders almost immediately, but the CPUC chose rulemaking over enforcement — the gap that had saved Uber in 2010 now protected its competitors. Once tolerance was clear, Uber moved in two steps: UberX launched in July 2012 as a cheaper tier using smaller licensed vehicles, and in April 2013 — just as the CPUC finalized the rules creating a legal category for ridesharing — Uber opened UberX to non-professional drivers and announced, on its own blog, a company-wide policy of operating wherever regulators appeared unlikely to enforce. Paul’s verdict is four words: “UberX copied ridesharing, too.”
Uber did not pioneer regulatory defiance in ridesharing; it initially demanded regulatory enforcement. What it did better than anyone was fast-follow at overwhelming scale once someone else had absorbed the legal pioneering risk. Sidecar paid for the answer. Uber collected it.
Part V: The Wars
Regulation as strategy
From October 2010 onward, the regulatory doctrine was consistent: launch, absorb the cease-and-desist, mobilize riders as a political constituency, negotiate from occupation. By 2014, “principled confrontation” reportedly appeared by name in the company’s internal values list, with Kalanick’s reported framing that defying regulators “becomes a negotiation.”
It worked because agencies could issue orders faster than they could enforce them, and every enforcement attempt made Uber more popular with the voters regulators answered to. But instrumentalized law-breaking has a terminal form, and Uber built it: Greyball, software that identified suspected regulators and showed them a fake version of the app — in the S-1’s own language, “a tool to limit the vehicle views available to regulatory enforcement authorities.” The New York Times revealed the program in March 2017; by May the Department of Justice had opened a criminal investigation. The distance from publishing a cease-and-desist on your blog to deploying software that deceives law enforcement is the distance this company traveled in seven years, in a straight line.
China: the war the money couldn’t win
Uber entered China in earnest in February 2014, into a market already militarized: the 2013–14 Didi–Kuaidi subsidy war had reportedly burned around $700 million combined before the two merged in early 2015. Uber’s campaign became the largest capital burn in startup history to that point — Kalanick himself reportedly acknowledged losing roughly $1 billion a year there, about $2 billion in total. The war was symmetric; the resources were not. Didi was better funded at home, and the formula that crushed fragmented, undercapitalized American taxi fleets failed against an equally financed, equally determined opponent. By Q1 2016, one analysis put Didi at 85.3% of China’s private-car hailing market against Uber China’s 7.8%.
The exit came on August 1, 2016 — days after China legalized ride-hailing by regulation. Uber sold Uber China to Didi in a deal valuing the combined entity at $35 billion (Uber China contributing roughly $7–8 billion by contemporaneous accounts), with Didi separately investing $1 billion in Uber at a reported $68 billion valuation and Kalanick taking a Didi board seat. On the stake, trust the audited source: the S-1 states Uber received “an approximate 18.8% interest” in Didi at closing, valued at approximately $6.0 billion, which Uber estimated had diluted to about 15.4% by September 30, 2018 — the commonly cited flat “18%” is a rounding of earlier reports. The arithmetic is remarkable: Uber lost roughly $2 billion fighting Didi and exited holding paper worth roughly $6 billion — a defeat that out-earned most victories.
2017: the machine turns on the operator
The collapse ran barely five months.
January 2017. Accused of trying to profit from airport protests over the new executive order on immigration, Uber watched #DeleteUber go viral. The S-1 quantifies the damage: “hundreds of thousands of consumers stopped using the Uber platform within days of the campaign.” The company that weaponized consumer love against regulators discovered the weapon fired both ways.
February 19, 2017. Susan Fowler, who had joined Uber as a site reliability engineer in November 2015, published “Reflecting on One Very, Very Strange Year at Uber.” Per her account: on her first official day on her new team, her manager propositioned her for sex over the company chat system; she took screenshots and reported him to HR immediately; HR declined to punish him because he was “a high performer” and it was supposedly his first offense — a claim that collapsed when she later met other women who had reported the same manager before her. She described an organization whose women had fallen from over 25% of her group to under 6%, and a performance score retroactively changed after calibration. The post was precise, factual, and unanswerable. Kalanick responded — “There can be absolutely no place for this kind of behavior at Uber” — and hired former U.S. Attorney General Eric Holder to investigate.
February 28 – March 3, 2017. Dashcam video surfaced of Kalanick berating an Uber driver over falling rates; he apologized, saying he needed to “fundamentally change as a leader and grow up.” Three days later the Times published Greyball.
June 2017. On June 11 the board unanimously adopted every Holder recommendation, including reducing Kalanick’s authority. On June 13 Kalanick announced an indefinite leave — writing that he needed “to work on Travis 2.0 to become the leader that this company needs” — and board member David Bonderman resigned the same day after a sexist remark. On June 20, investors including Benchmark — the firm behind the canonical bull case — forced the issue, and Kalanick resigned.
The standard telling treats 2017 as a culture scandal interrupting a growth story. The record supports a harder reading: the growth machine and the culture were the same machine. “Principled confrontation” produced both the city playbook and Greyball; the refusal to flinch produced both the China war and the HR department Fowler described. You could not fire the culture and keep the machine, so the investors fired the operator and spent two years paying down both.
The audited bill arrived with the S-1: operating losses of $4.0 billion in 2017 and $3.0 billion in 2018, an accumulated deficit of $7.9 billion by December 31, 2018 — including the one-time gains from selling the China and Russia operations for equity. Horan’s own tally, methodology disputed but direction not, puts cumulative losses above $20 billion. That is what the growth machine cost.
Structural Analysis: The Thinking Tools
Two frameworks earn their keep as thinking tools — they organize the situation; they don’t judge it.
Porter’s Five Forces, run on the 2010 taxi industry, explains the invasion’s speed. Rivalry among incumbents was administratively suppressed — fixed fares, capped supply — so none had ever competed on product. Supplier power (drivers) was low and about to collapse. Buyer power was atomized. Substitutes — street hails, car ownership — were the real market, which almost everyone missed. And the threat of new entrants was held at zero by one force: regulatory enforcement. Uber’s founding insight, proven in October 2010, was that the barrier was made of enforcement capacity, not law — and enforcement capacity was a bluff.
Rogers’ Diffusion of Innovations explains rider-side adoption speed. Relative advantage over a street hail: enormous and instantly perceptible. Compatibility: no new habit — the app removed the phone call and the cash from an existing behavior. Complexity: one button. Trialability: engineered, via free launch rides and referral credits. Observability: the product performed itself in public, curbside. Near the ceiling on all five factors — rider adoption needed the subsidies far less than driver supply and competitive warfare did.
The Evidence Scorecard: Would You Have Funded Uber?
Now the judging tool — the gated, weighted scorecard described on the series hub, scored from the evidence of the time, not the answer key. Two snapshots: October 2010 (the month of the seed and the cease-and-desist orders) and June 2014 (the $17 billion Series D), with one deliberate exception flagged below.
The gates, October 2010
Pain: real, frequent, urgent — inside a small set of cab-starved cities. Pass, narrowly. Buyer: the cleanest gate in the series — the rider pays, per transaction, card on file, above taxi rates, from day one. Pass. Market: on the period’s evidence — the founders’ own $4.2 billion sizing, the first investor’s ceiling — venture scale required believing something no document supported. Near-fail; the entire story lives in this gate. Behavior change: nearly zero for riders (the product deleted steps from an existing behavior), moderate but incentive-aligned for drivers. Pass.
Snapshot 1 — October 2010: 52/100
| Category (weight) | Score | The evidence at the time |
|---|---|---|
| Product-market fit (30) | 17 | Paying repeat usage at premium prices in one city; organic pull (the seed investor found it via tweets). Thin, but unambiguous. |
| Distribution (25) | 12 | Word of mouth in one dense city; no second city, no playbook, no repeatable channel demonstrated. |
| Unit economics (20) | 12 | A commission on premium rides, no fleet, no subsidies yet. Plausible; “profitable by design” still narrative. |
| Market quality (10) | 4 | Every contemporary sizing said small; the “why now” (GPS smartphones) is most of the 4. |
| Team / founder-market fit (10) | 6 | One proven exit (Camp), one operator with documented scar tissue (Kalanick), a GM hired over Twitter. |
| Moat / defensibility (5) | 1 | None. VRIO fails at Rare and Inimitable — the app was copyable and would be copied. |
| Total | 52 |
Fifty-two sits below the 55 threshold: a trap unless the evidence was about to change. That is what the honest evidence said — and how the smart money behaved: the round was uncontested, Cuban and Greathouse passed, and Hayes, who said yes, wasn’t underwriting the global company. Which surfaces something the scorecard doesn’t carry as a category: price. At $4 million, even the deck’s own $20-million-profit “realistic scenario” returns real money. At seed pricing, Uber was fundable on the small-market story alone. One round later it wasn’t.
Snapshot 2 — June 2014, with 2017 priced in: 67/100
One deliberate departure from the period-evidence rule: team and moat are scored on the record through 2017 — the categories where the 2014 evidence was most misleading. The series caps team because retrospective halo is strongest there; in Uber’s case the correction runs downward.
| Category (weight) | Score | The evidence |
|---|---|---|
| Product-market fit (30) | 27 | SF revenue alone a “healthy multiple” of the ~$120M/year historic SF taxi/limo market (Kalanick, via Gurley); years of ~20%/month growth; riders paid surge premiums. |
| Distribution (25) | 19 | Playbook repeatable across continents (Paris city five in 2012; hundreds followed); docked because the channel was substantially purchased. |
| Unit economics (20) | 6 | Negative contribution margin in every contested market; China burning ~$1B/year; later audited at $4.0B and $3.0B operating losses (2017–18). |
| Market quality (10) | 9 | The one number that improved a hundredfold — by behavior, not narrative: a single city out-earning its own prior industry. |
| Team / founder-market fit (10) | 4 | Extraordinary execution speed — and the same decision pattern produced Greyball, Fowler’s account, the Holder report, and the ouster. |
| Moat / defensibility (5) | 2 | Liquidity network effects real but city-level and imitable: Lyft survived at home; Didi won China. VRIO: Valuable, not Rare, demonstrably Imitable. |
| Total | 67 |
Sixty-seven — interesting but unproven — at $17 billion. That unflattering number is the scorecard doing its job. The framework is non-compensatory: no category rescues a fatal weakness in another, and Uber at peak growth carried one — unit economics that never closed in any audited document of the era. What kept the company alive was not on this card; it was capital markets’ willingness to fund the gap for a decade. The move between snapshots (52 → 67) is real: PMF and market quality exploded. But note the two lines that fell — economics and team. PayPal, in this series, was broken economics repaired between snapshots; Uber is the inverse — proven demand, deteriorating economics, papered over by the largest private funding stream in history.
Kill criteria, stated as of October 2010
- If a major city successfully enforces a shutdown, the model dies locally and the precedent spreads. Fired repeatedly — 2010, 2012, the Greyball investigations — and never fully connected: enforcement capacity kept losing to occupy-and-negotiate.
- If the market really is the historic ~$4B taxi/limo pool, venture returns above seed price are impossible. The criterion every rational passer relied on. It fired in reverse: the market expanded instead of binding.
- If an equally capitalized competitor contests supply, the subsidy spiral never lets economics close. Fired twice — partially at home (Lyft survived), completely in China (Didi won).
- If growth requires permanent subsidy — if contribution margin never turns — the company is renting its market. Fired, and never un-fired in this era.
- If the confrontation culture generates a scandal that reaches demand or decapitates leadership, the machine seizes. Fired in full sequence in 2017: #DeleteUber, Fowler, Greyball, Holder, ouster.
The kill criteria read as the wars section’s table of contents. Four of five fired; the company survived because criterion two fired in its favor hard enough to keep capital coming while the others burned.
What the Instrument Cannot See
Here is where the framework audits itself.
Every serious, evidence-based sizing of Uber’s market was wrong, in the same direction, for six years. The founders’ deck: $4.2 billion. The first institutional investor: “bigger than the black car market,” but no further. In June 2014, Aswath Damodaran published a rigorous analysis pegging Uber’s value around $5.9 billion, built on the assumption that its addressable market was the historic global taxi and car-service market, which he sized at $100 billion. On July 11, 2014, Gurley published his rebuttal, “How to Miss By a Mile,” and its decisive evidence was behavioral: Kalanick had claimed that June that “San Francisco total spend on taxi and limo was like 120 million bucks” — and that Uber’s SF business alone was already “a very healthy multiple bigger than that… it’s not about the market that exists, it’s about the market we’re creating.” (The boast almost certainly meant gross bookings, not net revenue.) Gurley’s expansion math — a car-for-hire market 3–6x the historic one, plus partial substitution of a car-ownership cost pool he put at $6 trillion globally — produced “a potential range of new TAM estimates from $450 billion all the way up to $1.3 trillion.” Roughly 100 to 300 times the founding deck.
The methodological lesson is about created markets. Bottom-up sizing — the exact discipline my market-quality category rewards — prices the market that exists. It cannot price a market the product calls into being, because there is no denominator to measure. When demand evidence starts violating the ceiling of every credible sizing, the correct response is not to average the estimates; it is to throw out the denominator and re-underwrite. Damodaran read the instrument correctly and missed. Gurley noticed the instrument was broken and updated. Applied mechanically in 2010, my scorecard keeps you out of Uber — not because it is wrong, but because it has a known failure mode: it underprices companies whose product expands the category. That failure mode is rare. When it’s live, it’s everything.
Hidden Forces Nobody Talks About
The incumbent’s weakness was regulatory, which made it political. Taxi fleets couldn’t cut prices (fixed by commission), add supply (capped by medallion), or improve dispatch (no capital, no engineers). Their entire defense was enforcement — and the CPUC, by the academic account, was short on staff and never pressed its 2010 demands, while the 2010 backlash showed enforcement carries political cost. Every unenforced order taught the market — including Uber’s future competitors — that the true price of defiance was approximately zero.
Sidecar and Lyft were Uber’s unpaid legal R&D. The peer-to-peer model that became UberX — essentially all of Uber’s eventual scale — was invented by others, under donation-model legal cover, while Uber demanded those inventors be shut down. The pioneers absorbed the regulatory risk; the fast follower with the war chest collected the category. Sidecar is barely a footnote now, which is the point.
Private capital replaced the public market’s referee function. In any prior era, a company burning at Uber’s rate would have needed public markets years earlier — and faced the unit-economics conversation in 2013 rather than 2019. ZIRP-era late-stage capital — founder-dictated terms, ratchets in place of diligence, a celebrity wiring extra millions uninvited — deferred it for a decade, and the S-1 is where the deferred questions got answered all at once.
The margin story underneath was labor pricing. Horan argues driver take-home pay fell more than 40% after Uber entered markets, and that unilateral pay cuts supplied most of the pre-IPO margin improvement. That is the critic’s framing — but a meaningful share of what looked like software-margin improvement was the repricing of drivers who had no collective counterparty.
The Luck Audit
Specific breaks, and the skill that exploited each — neither alone explains anything.
Lucky: the GPS smartphone arrived exactly then. The idea dates to August 2008, weeks after the App Store opened, with GPS chips going ubiquitous over the following two years. The skill: recognizing before it was consensus that location-aware phones turned dispatch into software, and building immediately.
Lucky: the recession manufactured driver supply. Idle black cars in 2010, then millions of underemployed people with cars through the UberX era. The skill: the incentive stack and cold-call groundwork that actually harvested that supply, city after city, faster than anyone else.
Lucky: the capital regime funded a decade of losses. No prior startup had been permitted to lose billions annually while private; ZIRP made the money available. The skill: fundraising as an offensive weapon — raising ahead of need, setting terms, converting balance sheet into market share faster than competitors could match. China is the control case: against equal capital and will, the same skill produced a $2 billion loss, cushioned by an exit window that opened days after Beijing legalized the industry.
Lucky: rivals de-risked the pivotal product. Sidecar and Lyft ran the peer-to-peer legal experiment first, and regulators blinked at them first. The skill: recognizing within months that the tolerated model obsoleted black cars, and scaling it through an already-built launch machine before the pioneers could fund one.
And the deepest luck: their own forecast was wrong in the good direction. The founders undersized their market by two orders of magnitude and were saved from their own spreadsheet; most TAM errors run the other way and kill quietly. The skill: noticing the error early — Kalanick was publicly re-underwriting his own market by 2014 — and pricing every round as if the corrected number were true.
What This Actually Means
I’ll resist the lessons-for-founders ending — most of this configuration isn’t replicable. But three patterns are worth holding together.
Demand was real; the economics were rented. Gurley’s bull case was right about the network effect, the market expansion, and the behavior that had already falsified the skeptics’ denominator. Horan’s bear case was right that the growth was purchased below cost, the moat thinner than the valuation implied, and the audited numbers coming. Both were right simultaneously for a decade. Any account of Uber that resolves that tension is discarding evidence.
The machine and the culture were one artifact. The playbook that opened five hundred cities and the software that deceived police were built by the same doctrine, staffed by the same filter, praised by the same board — until 2017, when the costs routed through demand, leadership, and the cap table at once. Build a company whose core competence is not flinching, and expect it, eventually, not to flinch at you.
Every instrument has a failure mode; know yours. My scorecard, honestly applied in October 2010, says 52 and pass — and the round’s actual buyers behaved accordingly. The instrument wasn’t misread; it was blind in a nameable place: markets the product creates rather than enters. The discipline is to watch for the tell that it’s failing.
The honest summary: a mediocre-scoring 2010 deal, correctly passed on by rigorous people, became one of history’s great venture outcomes because the gate everyone scored lowest was miswired in the company’s favor — and the company converted that break, with operational brilliance and other people’s money, into a global machine whose audited economics still hadn’t closed when the era ended. The people who called it a miracle and the people who called it a mirage were, for ten straight years, both looking at accurate data.
Sources and Notes
Primary sources:
- Uber Technologies, Inc., Form S-1 (April 2019), SEC EDGAR — the incorporation chronology, operating losses and accumulated deficit, the Didi stake, the #DeleteUber quantification, the Greyball acknowledgment.
- Garrett Camp, “The Beginning of Uber,” Medium (August 23, 2017) — the founder’s dating of the idea and the republished deck.
- Bill Gurley, “How to Miss By a Mile” (Above the Crowd, July 11, 2014) — the TAM range and the quoted Kalanick remarks.
- Susan Fowler, “Reflecting on One Very, Very Strange Year at Uber” (February 19, 2017).
- Kevin Novak’s first-person account of building surge pricing (Substack); Sunil Paul, “The Untold Story of Ridesharing, Part III” (Medium).
- TechCrunch, “Ubercab, Now Just Uber, Shares Cease And Desist Orders” (October 25, 2010), with the orders themselves.
Secondary and documentary sources: Brad Stone’s The Upstarts (2017); Forbes’ 2019 review of the 2008 deck; Forbes and TIME on the 2016 Didi deal; Fortune’s 2017 collapse timeline; CNN Business’ interviews with the earliest investors; the Harvard GSD account of the California regulatory sequence; the 20VC interview with Rob Hayes; John Greathouse’s account of his pass; Hubert Horan’s “Can Uber Ever Deliver?” series.
Disputed or single-source details:
- Surge pricing origin — the late-2011 dating and throughput claims rest on Kevin Novak’s own account; one retrospective instead says New Year’s Eve 2010. I privilege the first-person account.
- Driver incentive dollar figures — the referral bonuses come from a single early-employee account, the hourly guarantees from retrospective accounts; directionally consistent, individually unverified.
- The 2010 cease-and-desist particulars — the October 20 date and the penalty threats are from the orders as republished contemporaneously; no second confirmation of the amounts.
- Kalanick’s Red Swoosh exit (~$19M, 2007) — widely reported, not confirmed in a primary document cited here.
- Round valuations — the $4M seed and ~$60M Series A are “reported” figures. Didi’s 2016 investment in Uber is variously put at a $68B or $62.5B valuation; Uber China’s value in the merger at $7B or $8B.
- The Didi stake — contemporaneous reporting described 5.89% of shares carrying a 17.7% economic interest; the S-1 states approximately 18.8% at closing. I anchor on the S-1.
- Expansion pace — an ex-employee recollection of “a city a day” and 500 cities by end-2014 conflicts with contemporaneous statements; treated as memory inflation. Contemporaneous China ride-volume figures conflict even more wildly, so I lean on the Q1 2016 market-share numbers.
- Investor passes — Cuban’s is secondhand and hedged; Greathouse’s is his own account. Hayes remembers his outreach email as four words; CNN records it as “I’ll bite.”
- Camp’s $800 New Year’s Eve bill is from his own telling; the ubercab.com registration date (reportedly August 4, 2008) is from profiles of Camp and Stone’s reporting.
- Cumulative losses above $20 billion is Horan’s calculation, not an audited figure; “more than $5 billion raised by 2015” is from contemporaneous retrospectives, not a filing.
Analytical frameworks:
- The gated, weighted evidence scorecard (gates → weighted categories /100 → two snapshots → non-compensatory rule → kill criteria) is this series’ standard judging instrument, described in full on the series hub; the PayPal piece used its earlier four-axis version.
- VRIO derives from Jay Barney’s resource-based view of the firm (Journal of Management, 1991). Porter’s Five Forces is a thinking tool here, not a judging tool. The adoption lens is Everett Rogers’ Diffusion of Innovations. The team cap follows the hub’s reasoning, per Gompers, Gornall, Kaplan, and Strebulaev (Journal of Financial Economics, 2020).
Characterizations inferred rather than directly sourced are flagged in the text with “reportedly” or similar qualifications.