Google: The Origin of Search

In one sentence: when Google was founded in 1998, a dozen search engines and portals already existed — it wasn't first to the market, it was last. Two Stanford PhD students practically gave the technology away for free: Excite's CEO killed a million-dollar license on the spot because better search would send users away faster; Yahoo and Infoseek turned it down too. With nowhere left to sell it, they went back to school and built the company themselves. Twenty-seven years later, Alphabet generates more net income than any other US company — more than Apple, Microsoft, ExxonMobil, JPMorgan Chase, Berkshire Hathaway — runs an 87% gross margin on core search, and in 2025 was ruled a monopolist by the US government with 90% market share. Acquired's thesis: Google didn't win because it had the smartest algorithm. It won because it welded "the best technology" onto a distribution machine and an auction engine nobody wanted to copy and nobody who tried could catch — from the AOL bet that nearly bankrupted the company to a toolbar that rode into your computer as a stowaway inside Adobe and Dell installers. In David's words: "It is the most magical economic transaction I think ever known man."

The Company on One Page

Founder Profiles

Larry Page

His pedigree was itself a head start: both parents held CS PhDs from Michigan, his mother taught programming at Michigan State, and his brother Carl Jr., nine years older, also studied CS before heading west to Microsoft and Mentor Graphics. At six, he spent a year in Palo Alto during his father's Stanford sabbatical — Silicon Valley entered his childhood two decades before most founders ever see it. Ben and David open the episode determined to correct a persistent myth: Larry and Sergey were not accidental geniuses who stumbled into a great business. The ambition was there from the start. In Larry's own words:

"Probably when I was 12, I knew I was going to start a company eventually. I wanted to make the world better. In order to do that, you need to do more than just invent things."

His dissertation began as a system for annotating web pages, a decentralized alternative to Yahoo's human-curated directory; stuck on the question of who gets to judge whether an annotation is credible, he pivoted toward ranking itself — a turn he later summarized in a line the hosts return to again and again:

"It wasn't that we intended to build a search engine. We built a ranking system to deal with annotations."

When nobody would buy BackRub, his response wasn't discouragement — it was to walk the project back to Stanford and build the company himself: "We said whatever. We went back to Stanford to work on it some more." That same stubbornness resurfaces almost identically in 2002, when Google bets the entire company on the AOL deal. Told they were risking everything, his answer was: "We should be able to monetize the pages. If not, we deserve to go out of business." He held something close to an allergy to advertising — insisting early ads stay text-only, partly on taste, more firmly because images would slow the page down.

Sergey Brin

His path started from the opposite extreme: a Jewish family in Moscow, sharing a three-room apartment with his grandmother, his father an extraordinarily talented mathematician. Around 1977–78, his father attended an international math conference and came home determined to get the family out of the Soviet Union — it took two years to secure permission to leave. In America, his father became a math professor at Maryland, his mother a researcher at NASA Goddard. Sergey was precocious by any measure: high school at 16, a double major in math and computer science at Maryland finished in three years, graduating at 19, and an internship at Wolfram Research (the makers of Mathematica) the summer before Stanford. He and Larry met during prospective-student weekend and were sparring partners and soulmates from the first night — they shared an office at Google for years afterward. Where Larry ran more polished and product-focused, Sergey ran wilder: he drew Google's original rainbow logo himself in the open-source tool GIMP. Ben and David make a point of how rare a truly equal co-founder partnership like this is on the show — Bill Gates and Paul Allen quickly sorted into a clear hierarchy, NVIDIA's Jensen-Curtis-Chris trio eventually collapsed into just Jensen, and the closest parallel they can find is Tom Murphy and Dan Burke at Capital Cities. Larry and Sergey's "one plus one equals a hundred" partnership held for decades.

The Playbook

Each entry: origin story → insight → effect.

1. PageRank was born from an abandoned idea

• Story: Larry's original dissertation topic was a web annotation system; stuck on who gets to judge which annotations are credible, he pivoted to ranking itself, borrowing academia's own logic that citation count equals authority — backlinks are citations, anchor text is free metadata.
• Insight: the sharpest insights often don't arrive by aiming straight at the goal — they turn up while you're stuck on the sub-problem blocking your original plan.
• Effect: BackRub is born, and two years later it's renamed Google.

2. The window was only ever open a crack

• Story: ranking by backlinks required crawling and storing a copy of the entire web, because HTML only exposes outbound links, never inbound ones. In 1996 that was barely feasible — the whole web had about 600,000 sites. A year or two later, indexing the same web would have cost tens or hundreds of millions of dollars, well past what a couple of PhD students could attempt.
• Insight: some opportunities have their window set by an external exponential curve — too early and the market doesn't need you yet (a human-curated directory was still good enough), too late and the cost has already outrun you.
• Effect: both hosts agree it was the perfect window — a few years earlier it would've been Yahoo, a few years later it would've been unaffordable.

3. Getting rejected is what forced them to found the company

• Story: in 1997, Larry and Sergey shopped BackRub for a license all over town — Infoseek and Lycos passed, Yahoo turned down a million-dollar offer, and a near-deal with Excite died the moment its CEO watched the demo and realized better ranking meant users would click through and leave his portal faster.
• Insight: an incumbent's own monetization model can blind it to — or make it actively hostile toward — a superior technology; this is counter-positioning working against Google first, and for it later.
• Effect: Larry's own summary — "We said whatever. We went back to Stanford to work on it some more" — being rejected by the market is what turned this into a founding story.

4. Raise money on a plan you know is probably wrong, just to buy time

• Story: the Series A pitch deck, written by employee #9 Salar Kamangar, led with enterprise search licensing as the primary revenue driver, CPM banner ads as a grudging second, and portal search backfill as a third — in hindsight, the auction model that actually built the company didn't exist yet.
• Insight: sometimes a raise runs on a plausible-but-wrong plan, and its job isn't to lock in the eventual answer — it's to buy enough runway to find it.
• Effect: Sequoia and Kleiner Perkins jointly lead a $25M round that funds the company through to the moment the real model gets built.

5. Distribution is bought, not waited for

• Story: the Google Toolbar, shipped December 2000, gets bundled into installers for Adobe, RealNetworks, and WinZip, plus a direct deal to preinstall on Dell PCs. A toolbar user searches seven times as often as a non-toolbar user; average annual revenue per user jumps from $2 to $10-plus — that gap is the budget for buying distribution.
• Insight: once you know exactly what a user is worth, what you're willing to pay for distribution stops being a guess and becomes a computable ceiling.
• Effect: the toolbar strategy later becomes the default-search deal with Firefox (Mozilla's main revenue source for decades), a bundle inside Google Earth, and eventually the multi-billion-dollar traffic-acquisition payments to Apple for default placement in Safari.

6. The auction gets better with scale, not just cheaper

• Story: Ben's on-the-spot "quintessence" — ordinary scale economies just lower unit costs; Google's auction model also raises revenue per unit as it scales, because more bidders on a keyword means better price discovery and a higher clearing price, and a deeper pool of advertisers means more searches get monetized at all.
• Insight: this is a kind of "super scale economy" that doesn't cleanly fit even Hamilton Helmer's 7 Powers framework — David tries to file it under network economies, but both hosts admit live that the label doesn't quite hold.
• Effect: every search is worth more to Google than to any smaller rival, so Google can always outbid everyone for distribution — once the flywheel is spinning, it's brutally hard to catch.

7. Leave money on the table, on purpose, forever

• Story: rebuilding AdWords in 2001, Google adopts a Vickrey second-price auction — the winner pays only a penny more than the second-highest bid. Sheryl Sandberg, struggling to explain it to advertisers, calls her mentor and former Treasury Secretary Larry Summers, who tells her it's the same mechanism the Federal Reserve uses to sell Treasury bonds.
• Insight: not squeezing maximum value out of every single transaction buys advertisers' trust and spares them from constantly re-checking and fiddling with their bids — a store of long-term potential energy purchased with short-term revenue.
• Effect: Overture had discovered the same click-through-rate signal earlier but abandoned it because it broke their transparent pricing page and confused advertisers — handing Google the more durable mechanism for free.

8. Bet the company on a certainty, not a hope

• Story: the 2002 AOL deal — an 85% revenue share and a $100M revenue guarantee, signed while Google doesn't have $100M in the bank. Sergey's words: "We could have gone bankrupt. This is quite literally Google betting the company."
• Insight: the bet wasn't reckless — it rested on a fact they'd just proven. Once "Project Sunset" moved every advertiser onto the new auction model, revenue jumped from $86M to $440M in a single year; Google already held a working, repeatable model, and just needed more inventory to run it against.
• Effect: AOL brings in $35M in the first half alone and $200M the following year, blowing past the guarantee, and Google becomes a first-tier paid-search player almost overnight, taking share directly from Overture, AOL's prior partner.

Moat Analysis (the 7 Powers framework)

The episode's timeline stops around the 2004 IPO. The hosts are explicit that Switching Costs, Process Power, and Cornered Resource "aren't there yet" and belong to later chapters of the story.

Bear Case

Recorded in mid-2025, the episode opens by noting that Google's "seemingly unassailable" position has faced its first real challenge only this year, as the AI wars heat up. Alphabet's P/E ratio sits at an all-time low of 20, well below Apple (30), Microsoft (37), Meta (27), and NVIDIA (46) — even though its revenue is growing as fast as or faster than every peer except NVIDIA, the market is plainly pricing in the risk that AI upends search, and David can't resist noting the irony that this is the same company that published the Transformer paper. Looking back, the company's history wasn't a straight climb, either: it had essentially no real revenue through the dot-com winter, surviving on Yahoo's backfill deal and investment, and the AOL deal wasn't a thought experiment about risk — it was a real bet-the-company moment that could have gone the other way.

Bull Case

Once the auction-liquidity flywheel is spinning, it's extremely hard to catch — bigger scale lowers unit cost and raises unit revenue at the same time, a compounding structure that reinforces itself. The dual-class share structure kept Larry and Sergey in control, letting them make non-consensus calls — the AOL bet, the Dutch-auction IPO — that an outside board would likely have vetoed. Core search runs at an 87% gross margin, and in 2025 the US government formally ruled Google a monopolist with 90% market share — a legal liability, yes, but also hard proof the moat is real. Even at the height of the dot-com bubble's hottest talent market, the company recruited generational engineers like Urs Hölzle and Jeff Dean — the compounding of people and culture was already working in its favor.

Deep Cuts

• The GoTo/Overture tragicomedy: Bill Gross pitched pay-per-click plus bid-ranked results on stage at TED in 1998 and got booed. Because he'd disclosed the idea publicly more than a year before trying to patent it, the company missed the filing window and never locked down the core mechanism. GoTo rebranded as Overture, built a business worth well over $100M a year white-labeling paid search for AOL, MSN, and Yahoo (the Yahoo contract alone eventually accounted for roughly 75% of Yahoo's total revenue), and was later bought by Yahoo for $1.6B after a bidding war with Microsoft — but before Google even closed its Series A, Larry and Sergey had offered to sell Google to GoTo for $200M, and Overture's board turned it down rather than dilute their own $2B company by 10% for a startup with no revenue. Possibly the most expensive conservative board decision in history.
• Yahoo's CEO misjudges the room: at a dinner in late 2001, then-Yahoo CEO Terry Semel told Larry and Sergey that Yahoo — paying them under $10M a year as their biggest customer — didn't think they had much of a business, then offered to buy Google for $1B on the spot. He was turned down. A year later, after Google's ad revenue jumped 5x, Yahoo came back at $3B; Google countered with an offer to buy Yahoo for $5B instead — effectively a joke about reverse acquisition, and the last time the two companies seriously discussed a deal.
• Gmail and AdSense share a bloodline: engineer Paul Buchheit spent his 20% time from 2001 building a way to search your own mail, then prototyped matching ads from Google's own ad database against the content of your inbox. When Larry and Sergey saw it, they asked whether the same trick would work on ordinary web pages — directly leading to AdSense. Gmail itself launched on April Fools' Day 2004 with 1GB of free storage, dozens of times what Yahoo and Hotmail offered, and plenty of people assumed it was a prank.
• Sundar Pichai's first Google job: he joined in 2004 to run the client-products line that included the Toolbar — an unglamorous first assignment on the road to eventually becoming CEO.

Era & Industry Trivia

• AltaVista came out of DEC's research lab and was first to build a ten-million-page index using parallel crawling, but its parent company's real business was selling hardware, and it never treated search as a business worth running seriously.
• Yahoo was founded by two other Stanford PhDs, Jerry Yang and David Filo, and started life as a hand-curated directory — closer in spirit to a media company than a technology one.
• Early Google servers had no cases: motherboards were mounted directly onto corkboard and stuffed into racks, because the co-location facility charged by square footage rather than power draw, so engineers optimized ruthlessly for density; the whole fleet ran a hardware failure rate above 10% a year against an industry average of 3–4%, held together entirely by software-level replication.
• After buying Overture, Yahoo also bought Inktomi, trying to stitch together a Google competitor under the codename Project Panama — a footnote to that acquisition history is that WhatsApp founders Jan Koum and Brian Acton met while working at Yahoo, before getting fed up and leaving to start their own company.
• When Eric Schmidt joined in 2001, he discovered that half of Google's searches came from outside the US, yet there was no international ad sales operation — he half-jokingly told Omid Kordestani to get on a plane Monday morning and he'd call with the destination; within three years, international revenue climbed from 18% to 29% of the total.

Cross-domain Notes

This episode has no strong intersection with the PH discourse network — Silicon Valley company history and geopolitics are largely orthogonal here, unlike the TSMC episode, where Taiwan's chip geography sits at the very center of the PH domain. There's one thin thread worth flagging: the episode mentions, in a single line, Google's business in China between roughly 2002 and its 2010 withdrawal rather than continue censoring results — a concrete footnote to the US-China tech-rivalry theme that recurs throughout the PH domain's fortress-america and technate pages, though the episode doesn't develop it and leaves it for later chapters of the Google series. If there's a broader methodological echo, it's in the hosts' repeated focus on "founding windows" and "how a monopoly gets built" — structurally similar to the PH domain's interest in concentrated technological power — but that's a resonance at the level of analytical framework, not content, and shouldn't be overstated as a real intersection.

Pages Worth Creating

• Entities: larry-page, sergey-brin (founder pages: the Stanford meeting, PageRank's origin, "nobody would buy it so we built it ourselves"), eric-schmidt (the third leg of the triumvirate, Novell/Sun background)
• Concepts: 7 Powers 护城河框架 (Hamilton Helmer's framework, the standard analytical tool across the Acquired series; this episode contributes the "auction-driven super scale economy" as a variant the framework doesn't precisely name), pagerank (the episode's core algorithm, worth its own concept page tracing its technical evolution)
• Episode cross-links: Amazon.com：从网上书店到万物商店 (Jeff Bezos as Google's angel investor), NVIDIA 之一：GPU 公司（1993-2006） (NVIDIA's three-way founder partnership collapsing into just Jensen, a contrast to Larry and Sergey's decades-long equal partnership)

Source · acquired