Google

The 'botched' IPO that returned 16,000%: Wall Street laughed at the auction, then watched Google compound for two decades

In August 2004, the consensus was that Google had fumbled its debut. The modified Dutch auction was mocked by the banks it disintermediated, the price range was slashed from an initial $108-135 to a final $85, the share count was cut, and a Playboy interview with the founders — published as the roadshow ran — nearly tripped the SEC's quiet-period rules and had to be bolted onto the prospectus. Google priced on August 19, 2004, raising about $1.67 billion at a valuation just over $23 billion. Then the 'botched' deal closed up 18% on day one, and the lowest monthly close in the entire 22-year record — reached twelve days after the IPO — was still 52.3% above the offer price. The stock never looked back: as of August 2026 it stands at +16,194% versus the IPO, having peaked at +17,990% in March 2026.

The filing's self-forecast of decelerating growth proved almost comically premature, because the company spent the next decade buying and building its way out of every structural risk it had listed. The non-PC-device risk — the most prescient paragraph in the S-1 — was answered by acquiring Android in 2005, turning the mobile platform shift from an existential threat into a distribution moat. YouTube ($1.65 billion, 2006) and DoubleClick ($3.1 billion, 2007) extended the ad machine into video and display. Microsoft's integrated-search offensive and Yahoo's portal never dented search share, and the 2015 reorganization into Alphabet formalized what the Owner's Manual had promised: a founder-controlled conglomerate free to fund risky bets from a search-ads annuity. When Larry Page and Sergey Brin stepped back from executive roles in December 2019, they remained exactly what the dual-class structure guaranteed — controlling shareholders.

The risks that did materialize were the ones the S-1 could only gesture at. The 'unsettled laws' paragraph matured into the defining antitrust case of the internet era: on August 5, 2024, Judge Amit Mehta ruled that Google had illegally maintained a monopoly in general search, a case built partly on evidence that Google paid Apple roughly $20 billion a year for default placement. Yet even here Google's luck held — the September 2, 2025 remedies decision rejected divestiture of Chrome or Android, ordering data-sharing and an end to exclusive default contracts instead, an outcome the market read as close to a best case. Meanwhile the S-1's eeriest line — competition from 'companies that are not yet known to us' — arrived in the form of ChatGPT-class AI answers, the first genuine threat to the search franchise since 2004. Google's response, Gemini and a reported $180 billion AI infrastructure program announced in February 2026, is the largest capital deployment in its history, and the stock's 2025-2026 surge suggests investors currently believe the incumbent wins again.

The governance experiment deserves its own verdict. Every warning in the S-1 about founder control came true precisely as written — public shareholders never got a meaningful vote, and the share classes multiplied further with the 2014 stock split that created non-voting Class C shares. It simply never mattered, because the controlled company outperformed nearly everything shareholders could have voted for. That is the awkward legacy: Google's results became the standing justification for every subsequent founder who demanded ten-to-one voting rights without a search monopoly to back it up.

Google arrives at its IPO as something almost unheard of for a 2004 internet offering: already large, already profitable, and openly contemptuous of Wall Street convention. The business is a two-sided advertising machine — AdWords sells keyword-targeted ads by auction, AdSense syndicates them across a network of partner sites — and roughly 95% of net revenues come from advertisers who can leave at any time. The filing pairs plain-English candor (risks name 'Eric, Larry and Sergey' by first name) with an aggressively founder-favoring structure: Class B shares carry ten votes to public Class A's one, and the shares themselves are sold through an untested modified Dutch auction whose own failure modes the company dutifully lists as risk factors. The bet offered to investors is simple: search is becoming the front door of the web, Google is the best at it, and the founders should be trusted to run it their way indefinitely.

If search keeps growing as the primary way people find information, Google's auction-based ad model converts that attention into revenue at extraordinary margins, and the AdSense network extends the machine across the entire web. The brand is already a verb, the hiring process is famously selective, and the engineering culture keeps shipping products (Gmail had just launched) that widen the moat faster than Microsoft or Yahoo can copy it. Dual-class control means the founders can keep investing in long-term, risky projects without quarterly-earnings pressure, and the company's profitability means it never needs Wall Street's goodwill — which is exactly how it priced the deal. In the best case, Google becomes the dominant gateway to the internet and compounds for decades under founder control.

With hindsight: The bull case was that search would become the internet's front door and Google would monetize it under founder control. Reality exceeded it on every axis: the company won search, then won mobile (Android), video (YouTube), display (DoubleClick), and browsers, compounding to +16,194% over the IPO price by August 2026 — roughly a 163x — while remaining founder-controlled throughout. Even the S-1's explicit warning of near-term growth deceleration and margin compression proved far too pessimistic.

• The business is scaling explosively, with growth constrained mostly by the company's own infrastructure and management capacity.

  “We have experienced, and continue to experience, rapid growth in our headcount and operations, which has placed, and will continue to place, significant demands on our management, operational and financial infrastructure.” source ↗

• The AdSense network extends Google's ad auction beyond its own site and is the engine of future revenue growth.

  “We consider this network to be critical to the future growth of our net revenues.” source ↗

• The Google brand — built with almost no paid marketing — is itself a durable competitive asset.

  “We believe that the brand identity that we have developed has significantly contributed to the success of our business.” source ↗

• An unconventional, innovation-first culture is the company's core competitive weapon.

  “We believe that a critical contributor to our success has been our corporate culture, which we believe fosters innovation, creativity and teamwork.” source ↗

• Extreme selectivity in hiring has concentrated engineering talent that rivals cannot easily replicate.

  “We have in the past maintained a rigorous, highly selective and time-consuming hiring process. We believe that our approach to hiring has significantly contributed to our success to date.” source ↗

1. Formidable competition from Microsoft and Yahoo

   didn't happen

   competition · structural

   Google names its enemies plainly: Microsoft plans to fuse search into Windows and can outspend Google indefinitely; Yahoo has bought Overture, Inktomi, AltaVista and AllTheWeb, and Google has just terminated its search deal with Yahoo. Either rival leveraging its platform could trigger 'a significant decline in user traffic.'

   “Both Microsoft and Yahoo have more employees than we do (in Microsoft’s case, currently more than 20 times as many). Microsoft also has significantly more cash resources than we do.” source ↗

   What happened: The named enemies never landed a blow on search. Microsoft's integrated-search strategy and later Bing spent billions without dislodging Google; Yahoo faded and eventually gave up on its own search technology. The real competitive threat came two decades later from AI companies the S-1 said were 'not yet known to us.'

2. Nearly all revenue comes from advertising

   didn't happen

   profitability · structural

   The entire company rests on one revenue stream — performance advertising — sold to customers with no contractual lock-in. If advertisers stop seeing returns, or shift budgets back to traditional media, the model has no second leg to stand on.

   “We generated approximately 95% of our net revenues in 2003 from our advertisers. Our advertisers can generally terminate their contracts with us at any time.” source ↗

   What happened: The concentration persisted — advertising remained the dominant revenue stream for two decades — but the feared advertiser flight never happened. The auction model kept proving its ROI, and the ad business funded cloud, hardware, and AI diversification from a position of strength.

3. Dual-class structure gives insiders control even below 50% ownership

   came true

   governance · structural

   Class B stock carries ten votes per share against one for the public's Class A. Larry, Sergey and Eric will control the company outright, public holders cannot meaningfully influence anything, and the filing concedes the company 'may take actions that our stockholders do not view as beneficial.'

   “In addition, because of this dual class structure, our founders, directors, executives and employees will continue to be able to control all matters submitted to our stockholders for approval even if they come to own less than 50% of the outstanding shares of our common stock.” source ↗

   What happened: Exactly as disclosed: the founders control Alphabet to this day, remaining controlling shareholders after stepping back from executive roles in 2019, with control extended further by the 2014 creation of non-voting Class C shares. Public shareholders never got a say — and never had cause to complain, given the returns.

4. Unsettled U.S. and foreign law governing online services

   came true

   regulation · structural

   Liability rules for search and hosted content are unresolved worldwide; European data-protection law is 'in flux' and could force changes to Google's data practices; regulated ad categories (pharmaceuticals, alcohol, firearms) create additional exposure. A business built on indexing everything and monetizing user data sits squarely in regulators' path.

   “The laws relating to the liability of providers of online services for activities of their users are currently unsettled both within the U.S. and abroad.” source ↗

   What happened: This became the defining risk of Google's maturity: European data-protection and competition regimes, and ultimately Judge Mehta's August 5, 2024 ruling that Google illegally maintained a search monopoly. The September 2, 2025 remedies decision spared Chrome and Android but ordered data sharing and banned exclusive default deals.

5. Revenue growth rates will decline and margins face downward pressure

   partly came true

   growth · serious

   Management explicitly forecasts decelerating growth and compressing margins as the revenue mix shifts toward lower-margin Google Network traffic, competition intensifies, and the law of large numbers bites. Rapid growth has also masked cyclicality and seasonality in ad spending that will re-emerge as growth slows.

   “We expect that in the future our revenue growth rate will decline and anticipate that there will be downward pressure on our operating margin.” source ↗

   What happened: Directionally inevitable, wildly premature. Growth stayed extraordinary for well over a decade after the warning, and margins never suffered the forecast compression. The honest self-deprecation that spooked 2004 investors turned out to be the cheapest expectation-setting in IPO history.

6. Search may miss the shift to non-PC devices

   didn't happen

   tech & security · structural

   The filing flags the migration of internet access to mobile phones, handhelds and set-top boxes, and warns that Google's PC-centric search may not translate — a candid admission that the company's dominance is tied to a device category it does not control.

   “If we are unable to attract and retain a substantial number of alternative device users to our web search services or if we are slow to develop products and technologies that are more compatible with non-PC communications devices, we will fail to capture a significant share of an increasingly important portion of the market for online services.” source ↗

   What happened: The threat was real and Google neutralized it decisively: it bought Android in 2005, which became the world's dominant mobile operating system and made Google search the default on billions of devices. The platform-shift risk became the platform-owning moat.

7. The auction may produce a 'winner's curse' and a rapid price decline

   didn't happen

   market · serious

   An entire risk section is devoted to the IPO mechanism itself: only a handful of auctions have ever been tried in the U.S., the underwriters' systems could fail, retail-heavy participation could increase volatility, and successful bidders may discover they were the highest fools in the room.

   “The auction process for our initial public offering may result in a phenomenon known as the “winner’s curse.” At the conclusion of the auction, bidders that receive allocations of shares in this offering (successful bidders) may infer that there is little incremental demand for our shares above or equal to the initial public offering price.” source ↗

   What happened: There was no winner's curse — there was a loser's curse for everyone who sat out. The stock closed up 18% on day one, and its lowest monthly close in 22 years of trading was still 52.3% above the $85 offer. The auction that Wall Street called botched delivered its buyers one of the great entry points in market history.

8. Reliance on Google Network members for a growing share of revenue

   didn't happen

   platform dependence · serious

   AdSense revenue depends on a small number of large partner sites, some of which compete with Google and could defect to rivals or demand a larger share of ad fees — and Google has guaranteed them minimums totaling roughly $544.8 million at March 31, 2004.

   “The net revenues generated from the fees advertisers pay us when users click on ads that we have delivered to our Google Network members’ web sites represented approximately 15% of our net revenues in 2003, and approximately 21% of our net revenues for the three months ended March 31, 2004” source ↗

   What happened: The predicted dependence inverted: Google's own properties (search, then YouTube after the 2006 acquisition) grew so fast that network revenue shrank in relative importance, and the feared partner defections never threatened the business.

9. Click-through fraud could cost advertiser confidence

   didn't happen

   operations · serious

   The pay-per-click model is exposed to fraudulent clicks; Google already regularly pays refunds and may owe retroactive ones. Undetected fraud degrades advertiser ROI and, at the limit, corrodes trust in the entire auction.

   “We are exposed to the risk of fraudulent clicks on our ads. We have regularly paid refunds related to fraudulent clicks and expect to do so in the future.” source ↗

   What happened: Click fraud remained a perpetual arms race and a periodic litigation topic, but detection systems kept it contained. Advertiser confidence in the auction never broke; the model became the standard for the entire industry.

10. Loss of Eric, Larry or Sergey

    didn't happen

    key person · serious

    The CEO and founders are declared critical to technology, culture and strategy; all executives are at-will with no key-person insurance, and — unusually — the initial option grants of senior management are already fully vested, so the money reason to stay has largely expired.

    “In particular, our CEO Eric Schmidt and our founders Larry Page and Sergey Brin are critical to the overall management of Google as well as the development of our technology, our culture and our strategic direction.” source ↗

    What happened: The trio stayed for the entire hypergrowth era, and every transition was orderly: Page took the CEO role back in 2011, the Alphabet reorganization installed Sundar Pichai atop Google in 2015, and when Page and Brin stepped back in December 2019 they remained board members and controlling shareholders. No key-person shock ever hit.

11. Intellectual property suits, led by Overture's AdWords patent claim

    partly came true

    legal · serious

    Yahoo-owned Overture claims the AdWords pay-for-placement model infringes its patent — a suit aimed at the heart of the revenue engine — while trademark-keyword cases in France, Germany and the U.S. multiply, amplified by Google's new policy of no longer disabling trademarked keywords.

    “For example, Overture Services (now owned by Yahoo) has sued us, claiming that the Google AdWords program infringes certain claims of an Overture Services patent.” source ↗

    What happened: The Overture suit was settled around the IPO for a block of Google shares, removing the threat to AdWords. IP litigation remained a constant of Google's existence — trademark-keyword cases, copyright fights over search and video, the decade-long Oracle Java war — but none ever crippled the core business.

12. New technologies could block Google's ads

    partly came true

    tech & security · boilerplate

    Ad-blocking software could strip out the paid links that generate most of Google's revenue — a short, almost throwaway paragraph flagging a threat to the only revenue stream the company has.

    “Technologies may be developed that can block the display of our ads. Most of our net revenues are derived from fees paid to us by advertisers in connection with the display of ads on web pages.” source ↗

    What happened: Ad blockers went mainstream, exactly as warned — and revenue grew anyway. Google hedged the risk structurally by building Chrome into the world's dominant browser, giving it influence over the very layer where blocking happens.

13. Corporate culture may not survive scale

    partly came true

    other · serious

    Google credits its culture for its success and warns that growth, bureaucracy, and the IPO's sudden wealth disparities among employees could erode it — an unusually introspective risk for an S-1.

    “As our organization grows, and we are required to implement more complex organizational management structures, we may find it increasingly difficult to maintain the beneficial aspects of our corporate culture.” source ↗

    What happened: The prediction aged well as a cultural matter: the famously open, mission-driven culture visibly strained at hundreds of thousands of employees — internal protests, ethics controversies, large layoffs. But the warned business consequence never followed; the machine kept compounding regardless.

14. Three-headed leadership may slow decisions

    didn't happen

    governance · serious

    By bylaw, the CEO and the two founder-presidents run the company as a committee. The filing admits this consensus model can stall key strategic decisions and that a falling-out among the trio would harm the business.

    “As a result, Eric, Larry and Sergey tend to operate the company collectively and to consult extensively with each other before significant decisions are made. This may slow the decision-making process, and a disagreement among these individuals could prevent key strategic decisions from being made in a timely manner.” source ↗

    What happened: The triumvirate held together for the decade that mattered and dissolved gracefully rather than in conflict — Schmidt to executive chairman, Page to CEO, then Pichai. The consensus-management structure the S-1 fretted about produced some of the best capital-allocation decisions in corporate history (Android, YouTube).

15. Rescission offer for securities-law violations in option grants

    didn't happen

    legal · serious

    Google issued shares and options under multiple stock plans without required registrations, forcing a rescission offer with potential liability of up to $34 million plus interest — sloppy legal housekeeping disclosed on the way to a mega-IPO.

    “Shares issued and options granted under our 1998 Stock Plan and our 2003 Stock Plan were not exempt from registration or qualification under federal and state securities laws and we did not obtain the required registrations or qualifications.” source ↗

    What happened: The rescission offer became a historical footnote: with the stock far above employees' exercise prices, taking the refund made no economic sense, and the up-to-$34 million exposure never materially mattered to a company of Google's scale.

How the S-1 reads

This is the filing that wrote the template half of this archive later copied: a profitable, founder-controlled company selling low-vote stock to the public and telling them plainly that they will have no say. The risk section is stylistically unlike anything before it — first-person, first-name ('Eric, Larry and Sergey'), and willing to disclose risks most issuers would bury: that the company's own IPO mechanism might hurt buyers, that its name might become a generic word, that its executives are already rich and fully vested. The structural tell throughout is concentration-by-design: one revenue stream (95% advertising), one governance bloc (ten-to-one voting), one auction to price it all — paired with a level of self-awareness about those concentrations that makes the document read less like a prospectus and more like a dare.

• “After our offering, our Class B common stock will have ten votes per share and our Class A common stock, which is the stock we are selling in this offering, will have one vote per share.” source ↗

  The dual-class structure this S-1 introduced to tech IPOs, stated in a single sentence.

• “For example, there is a risk that the word “Google” could become so commonly used that it becomes synonymous with the word “search.”” source ↗

  Google formally warns that its name might become a generic verb for searching — as a trademark risk.

• “In addition to Microsoft and Yahoo, we face competition from other web search providers, including companies that are not yet known to us.” source ↗

  A line that reads very differently two decades later, in the age of AI search.

• “For example, people have raised privacy concerns relating to the ability of our recently announced Gmail email service to match relevant ads to the content of email messages.” source ↗

  The first appearance of the privacy critique that would follow Google for twenty years, aimed at the weeks-old Gmail.

What this one teaches

• The market's verdict on an IPO's mechanics says nothing about the business. Google's auction was declared a failure, the range was cut from $108-135 to $85 — and the stock never closed a month below the offer price in 22 years. Process embarrassment and asset quality are independent variables.
• A company that names its real risks accurately can still be wildly underpriced. Google forecast its own growth deceleration and margin compression in plain language; investors took the candor at face value and missed a 163x.
• The best founders treat structural risks as acquisition targets. The S-1's non-PC-device warning became Android; the online-video copyright minefield became YouTube. Watch whether management buys its way ahead of the risk curve or merely discloses it.
• Dual-class governance is judged by outcomes, not principles. Google's ten-to-one structure delivered everything it promised shareholders and normalized founder control for a generation of far weaker companies — the archive's later filings copied the structure without the monopoly.
• The risk that finally bites is usually filed under 'unsettled law.' Competition from named rivals did nothing in twenty years; the vague regulatory paragraph produced a federal monopolization ruling in 2024. Boilerplate-sounding legal risks deserve more weight at monopoly-scale businesses.
• The vaguest competitive risk — 'companies that are not yet known to us' — is the one that eventually mattered. Two decades of dominance ended in genuine competitive uncertainty only when an unforeseeable technology shift (generative AI) changed what 'search' means.

The paper trail

1. 2004-04-29 S-1 filing index ↗ document ↗
2. 2004-05-21 S-1/A filing index ↗ document ↗
3. 2004-06-21 S-1/A filing index ↗ document ↗
4. 2004-07-12 S-1/A filing index ↗ document ↗
5. 2004-07-26 S-1/A filing index ↗ document ↗
6. 2004-08-06 S-1/A filing index ↗ document ↗
7. 2004-08-09 S-1/A filing index ↗ document ↗
8. 2004-08-11 S-1/A filing index ↗ document ↗
9. 2004-08-13 S-1/A filing index ↗ document ↗
10. 2004-08-16 S-1/A filing index ↗ document ↗
11. 2004-08-18 S-1/A filing index ↗ document ↗
12. 2004-08-19 424B4 filing index ↗ document ↗
13. 2004-10-26 S-1/A filing index ↗ document ↗
14. 2004-11-12 S-1/A filing index ↗ document ↗
15. 2004-11-18 S-1/A filing index ↗ document ↗
16. 2004-11-23 S-1/A filing index ↗ document ↗

Filed as GOOGLE INC.. All documents are public domain, served by SEC EDGAR.