The Class Of: The Y Combinator Summer 2007 Batch

The Million Dollar Question: Seven of the nineteen companies in Y Combinator’s Summer 2007 batch were acquired. For how many of those seven has the purchase price ever been publicly disclosed?
A) All seven B) Four C) One D) None

Read on for the answer.

Every other cohort in this series has a counting problem. The PayPal Mafia is a legend assembled from thirteen photographed men out of 776 who actually worked there. The WhatsApp 55 has a famous number that comes from journalism rather than a filing. The Google class of 2004 is the easy case only because the company was obliged to print its headcount in a prospectus.

Y Combinator’s Summer 2007 batch was supposed to be the clean one. Accelerator batches are published in advance. The companies stand up on a named day, in front of named investors, and somebody writes it down. The denominator should be knowable rather than inferred.

It nearly is. Query Y Combinator’s own company database for companies whose batch field reads Summer 2007 and it returns exactly nineteen. TechCrunch’s Demo Day report of 16 August 2007 agrees on the number: “a whopping 19 companies giving lightning fast 7 minute elevator pitches.”

The two nineteens are not the same nineteen. And the gap between them turns out to be the most honest thing about this batch.

The denominator is published, and it has two holes in it

TechCrunch named ten companies that day, noting that “some of the 19 companies declined mention in this roundup.” Six of those ten sit in Y Combinator’s directory under the same name. Two are there under names they took later — Fauxto, which the directory records as the former name of Splashup, and Adpinion, which was the product of a company the directory files as Hello Chair.

Two are not there at all.

Versionate was a collaborative document tool that TechCrunch, writing on 11 July 2007, introduced as “Y Combinator’s second company of the summer season, Versionate, launches today.” It exists nowhere in Y Combinator’s current index, under that name or as any company’s recorded former name. No founders, no acquisition, no shutdown note.

Cloudant — not the database company of the same name, which was Summer 2008 and went to IBM, but a hardware startup from this batch — is also missing. TechCrunch’s description is specific: “The Slapvid guys have changed their startup and come back as a hardware startup, Cloudant. Their router promises to take full advantage of your bandwidth by simultaneously downloading multiple parts of a file.” Neither SlapVid nor that Cloudant appears in the Summer 2007 list today.

So a reporter standing in the room named two companies by batch and by position that the batch’s own official history no longer contains. Either the directory has lost them, or the contemporaneous account was wrong. Either way, the cohort with the knowable denominator does not quite know its own denominator.

That is not an isolated glitch, and it is worth establishing early, because every number later in this piece that comes from Y Combinator rather than from a regulator carries the same caveat. The directory lists one founder for AppJet, David Greenspan — while the coverage of AppJet’s acquisition names three people, CEO Aaron Iba, CTO J.D. Zamfirescu and COO Daniel Clemens, none of whom is the one the directory lists. It files SocialPicks as “Inactive” when SocialPicks was acquired twice. Its own prose description of Hello Chair names three founders, one of whom its own founder field omits. Songkick is generally recorded as having been founded by three people; the directory lists all three, but third co-founders are exactly what these lists lose. The published roster is a primary source, and it needs fact-checking like any other.

The same roster does settle a great deal of misremembering. Justin.tv — later Twitch — was Winter 2007, not this batch. So were Weebly and Octopart. Heroku and Tipjoy were Winter 2008. Wufoo was Winter 2006. And the Cloudant that IBM bought in 2014 was Summer 2008, founded by three MIT physicists; it should not be credited to this cohort.

Two of the famous ones are genuinely here. Disqus is Summer 2007. So is Dropbox.

What $15,000 bought

Y Combinator’s terms in that era, as TechCrunch described them in March 2009: “Investments are small ($5,000 + $5,000/founder) in exchange for around 6% of equity.” The same piece, sizing how far a new Sequoia-backed fund would stretch, put the typical cheque at “an average investment of only about $15,000 per startup.”

A two-founder Summer 2007 company therefore got $5,000 plus $10,000, for roughly six per cent. That is the entire deal. Dropbox had two founders. Jessica Livingston, in Y Combinator’s own account, wrote the company’s first investment cheque on 26 June 2007. Fifteen thousand dollars for about six per cent implies a notional post-money valuation around a quarter of a million dollars — a back-of-envelope figure, not a priced round, but a useful marker of the altitude.

By 2010 the formula had crept to $11,000 plus $3,000 per founder for “a piece of pure equity of around 5%,” per Forbes. Today’s standard deal is $500,000: $125,000 converting into a fixed 7%, and $375,000 on an uncapped note. The 2007 cohort was funded at roughly three per cent of what a 2026 cohort gets, which is worth holding in mind before reading anything moral into how many of them failed.

One geographic detail matters more than it looks. Summer batches in that era ran in Cambridge, Massachusetts, and this batch pitched twice. Y Combinator’s own telling is that “Dropbox impressed investors in Boston. It just didn’t impress them enough to write a check” — and that Drew Houston and Arash Ferdowsi “presented in our Mountain View Office on August 16, raised money from Sequoia, and moved to San Francisco.” The batch’s defining outcome turned on which of its two demo days you count, in a programme geography that no longer exists.

One company in nineteen

Dropbox listed on Nasdaq on 23 March 2018, the first Y Combinator company to go public. The final prospectus priced 36,000,000 Class A shares at $21.00 — 26,822,409 sold by the company, 9,177,591 by existing stockholders.

A warning before any share count, because most secondary accounts get this wrong: Dropbox effected a 1-for-1.5 reverse stock split on 7 March 2018, between the first S-1 and the final prospectus. The S-1’s numbers are pre-split, the prospectus’s post-split. Compare them without adjusting and the founders’ stakes appear to grow between February and March. They did not.

Post-split, from the prospectus’s ownership footnotes, the shares each founder actually owned sat in trusts:

Drew Houston held 84,763,454 Class B shares across three — 8,781,548 in a remainder trust dated 30 December 2010, 75,481,406 in a revocable trust dated 7 September 2011, and 500,500 in a children’s trust dated 12 April 2012. At $21.00 those shares were worth $1,780,032,534.

Arash Ferdowsi held 34,270,718 Class B shares across two trusts. At $21.00, $719,685,078.

The percentages in the prospectus are larger than those holdings, and the difference is instructive. Houston is shown with 29.8% of shares outstanding and Ferdowsi with 12.2%, but those figures count unvested restricted stock and — more importantly — 16.9 million and 7.1 million Class B shares respectively that the founders did not own and merely held the votes to, under Co-Founder Voting Agreements. Together they controlled about 42% of the company going in and roughly 37.6% coming out the other side of the offering. What they owned was less than what they voted, which is the standard architecture and almost never reported as such.

Each of them sold exactly 2,333,333 shares into the offering. At $21.00 that is $48,999,993 gross, apiece — the same number to the dollar for both. Everything else was locked for 180 days.

Class B carries ten votes a share, and after the offering Class B represented approximately 98.1% of Dropbox’s voting power. Sequoia, which had come in after the Mountain View demo day, held 87,171,450 Class B shares — a quarter of the Class B stock and 23.2% of the company. Accel held 18,752,565, five per cent of the company.

Then there is the part almost nobody reports. In December 2017, before the IPO, Dropbox’s board granted both founders restricted stock vesting only if the share price held nine escalating targets over any consecutive thirty-day trading period. Houston’s grant covers 10,333,333 post-split shares in tranches at $30.00, $37.50, $45.00, $52.50, $60.00, $67.50, $75.00, $82.50 and $90.00. The expense Dropbox booked for Houston’s grant alone, at an estimated average fair value of $10.60 a share, was $109.6 million, fully recognised by the end of 2024.

Per Dropbox’s 2026 proxy statement, exactly one tranche has ever vested: 2,066,667 shares on 15 November 2021, at the $30.00 level. “The remaining tranches covering 8,266,666 shares remain outstanding… No additional tranches were earned or vested during 2025.” The window closes on 23 March 2028.

So Dropbox has expensed one of the signature founder mega-grants of the 2018 IPO class in full, and nineteen years after the batch it has paid out one time in nine. With the stock in the low thirties, the $37.50 tranche is within reach and the $90.00 tranche is a fantasy.

Ferdowsi resigned as an officer and director in March 2020. His 8-K quote is the best single line anyone from this batch has filed with a regulator: “The last 13 years have been incredible. We’ve come a long way since 2007 – from working out of a tiny apartment in North Beach to running a public company of nearly 3,000 people.” He has not appeared in an ownership table since, which means what he kept is no longer public. Forbes still carries a profile valuing him at $400 million — as of December 2016, pre-IPO, nearly ten years stale.

Houston stayed. As of 31 March 2026 he held 74,616,889 Class B shares: 98.99% of every Class B share outstanding, and 82.40% of the company’s total voting power, against roughly a third of its economics. Forbes put him at $2.4 billion on 26 September 2026.

Dropbox’s market capitalisation is lower than it was on its first day of trading — about $7.1 billion at $32.70 a share on 29 September 2026, against $11.2 billion at the 23 March 2018 close. That is not a loss for shareholders, and the distinction matters: the share price is above both the $21.00 offer and the $28.48 first-day close. The company shrank its own float, from roughly 392 million shares at the IPO to about 217 million, by buying its own stock back. Dropbox made $689.1 million of operating income on a 27.3% margin in 2025. One member of this batch is a profitable public company that has spent years handing cash back to its owners while its founder’s control has gone the other way, from 29% of the votes to 82%.

Seven acquisitions, zero prices

That is the answer to the question at the top: D, none.

Seven companies from this batch were acquired. Not one purchase price has ever been disclosed by the buyer, the seller, or anyone under an obligation to say.

Disqus went to Zeta Global, announced 5 December 2017. TechCrunch reported that “a source close to the two companies tells us that the acquisition price was close to $90 million” — a rumour, not a disclosure. I went looking for corroboration in Zeta’s own filings. Its 2021 Form S-1 never states a Disqus purchase price; the only Disqus line item is an acquisition-related liability of $6.451 million at the start of 2019, of which $4.649 million was paid during the year and $1.802 million written off as a gain on extinguishment. That is a holdback, not consideration. The $90 million figure has no primary-source trace anywhere.

Disqus is also, quietly, the most durable product in the batch. Zeta’s FY2025 annual report, filed in February 2026, names the DISQUS commenting system, alongside LiveIntent inbox advertising, as how the company collects individual data directly from consumers. Nineteen years after Demo Day, an S07 product is a named data asset in a public company’s 10-K.

AppJet, which built Etherpad, went to Google in December 2009. TechCrunch said the price was “apparently in the low eight figures”; Wikipedia says undisclosed. AppJet had raised about $700,000. Etherpad was open-sourced after the acquisition and is still in use, which makes it the second S07 product more widely used than its founders are rich. Aaron Iba, AppJet’s CEO, became one of Y Combinator’s full-time partners in 2011.

ZumoDrive — corporate name Zecter — went to Motorola Mobility on 22 December 2010; terms of the deal were not disclosed. It was a cloud storage company sold for an undisclosed sum three years before Dropbox was privately valued at ten billion dollars, by founders who had pitched in the same room as Dropbox’s.

Anywhere.FM was acquired by imeem at the end of 2007 — co-founder Sachin Rekhi’s own account is simply that “imeem acquired Anywhere.FM at the end of 2007,” which as far as I can establish makes it the fastest exit in the batch, within months of Demo Day. Then the acquirer died. MySpace agreed to buy imeem in November 2009 for a reported sum under $1 million and shut it down within days of closing. The quickest liquidity event in the cohort resolved to approximately nothing two years later.

Clickpass went to Synthasite in December 2008. TechCrunch’s whole account of the consideration is this: “There are no details on the purchase price or terms of the deal, other than that Clickpass CEO Peter Nixey will be joining the SynthaSite team.” Synthasite had raised $5 million; Clickpass had raised the Y Combinator seed.

SocialPicks was acquired by FinancialContent around April 2009 — price undisclosed — and again, by SocialInSight, in 2012, also for an undisclosed amount. Y Combinator’s directory lists it as “Inactive.”

Songkick is complicated enough to need its own section below.

Seven transactions, seven undisclosed prices, two press rumours. Only one company in this batch — Dropbox — has a number behind it that came from a filing. For the other eighteen there is no reliable figure that could go in a “what it was worth” column. Any table that fills those cells has invented them.

The ten that left nothing but a domain

Ten companies produced no exit at all. Y Combinator’s one-line description of ContestMachine is the batch’s epitaph: “Roll your own contest or giveaway. No longer in business.”

The others are Bountii, Biographicon, Eivod, Fuzzwich, Hello Chair, iJigg, Meetyu, Reble and Splashup. I fetched every one of their listed domains on 1 October 2026 and not one of them is a working product. Most return nothing at all — dead lookups, server errors, blank pages. Splashup is a for-sale placeholder. iJigg’s domain is listed for sale on a registrar’s marketplace. And anywhere.fm, whose company was the first in the batch to be acquired, now serves an Indonesian gambling page, while zumodrive.com redirects to a free wallpaper gallery. Three former Y Combinator companies are now domain inventory.

Rob Fitzpatrick, who co-founded Fuzzwich and later wrote The Mom Test, has given the most useful account anyone from this cohort has given of the experience. His company raised from top-tier investors, had Sony Music, MTV and the BBC as customers, and was “quietly dying in year four.” What he says he actually minded was not the failure: “The main thing I didn’t enjoy was feeling compelled to continue due to my obligation to investors and employees.”

Now the part that breaks the premise of this entire series.

Russ d’Sa is the founder Y Combinator records for Meetyu — status Inactive, team size zero, one founder. On 22 January 2026, d’Sa’s company LiveKit raised a $100 million Series C led by Index Ventures at a $1 billion valuation, and OpenAI uses it to deliver voice to ChatGPT. Rank the Summer 2007 founders by wealth created and second place belongs to the man whose S07 company Y Combinator scores as having zero people and zero outcome. The directory’s own founder bio for Meetyu reads “founder and ceo @ LiveKit.” The company record says nothing happened; the founder record, on the same page, points at a billion-dollar company.

Third place is similar. Lee Linden co-founded ContestMachine — the “no longer in business” one, which by his co-founder Savraj Singh’s account dissolved when Linden left for business school. Linden then built Karma, which Facebook acquired in May 2012 on undisclosed terms, and is now a venture investor at Quiet Capital. Singh himself came back through Y Combinator a second time, as the solo founder of Wattvision in Winter 2009 — his own Y Combinator bio reads “YC S‘07 and W‘09” — making him the one person who appears in two different batches from this story.

And one company from the batch is still operating, independently, under its original name. Blue Frog Gaming, founder Matt Maroon, is listed by Y Combinator as Active with a team of eleven. Its site now runs SFC2: Starfleet Commander Reborn, described there as “the rebirth of Starfleet Commander, the browser space MMO from the 2000s, rebuilt from the ground up by its original developer.” When I looked on 1 October 2026 it reported a few hundred commanders active that week, roughly 1,200 fleet missions under way and about 7,500 worlds under player control, under a 2026 copyright notice. No exit, no acquisition, no outside round, nineteen years.

Of nineteen companies, exactly two still operate under the name they pitched with. One is a Nasdaq-listed company worth about $7 billion. The other is a browser game with a few hundred players.

The two biggest numbers in the batch are not exits

Songkick’s story is the one a reader will most want simplified, and the one that most resists it.

In June 2015 Songkick merged with its competitor CrowdSurge in a 50-50 stock-for-stock deal alongside a $16 million Series C from Access Industries, Sequoia and Index. The combined company kept the Songkick name and ran with co-CEOs, founder Ian Hogarth and CrowdSurge’s Matt Jones. In January 2016 Jones became sole chief executive and Hogarth moved to chairman. In July 2017 Warner Music bought “selected assets” — the concert-discovery app and the company’s trademark, around 15 million monthly users — for terms that were not disclosed, explicitly excluding the ticketing business and the pending antitrust litigation against Live Nation and Ticketmaster.

Then, in January 2018, the remnant entity — reorganised as Complete Entertainment Resources Group — settled that suit for $110 million weeks before trial, with Live Nation also taking the ticketing platform and patent portfolio.

So the second-largest number attached to this batch is a litigation settlement, paid to a holding company, two years after the Y Combinator-era chief executive had stepped down, on a founding stake that had been halved by a 50-50 merger and then diluted by later rounds. It is not an S07 exit in any ordinary sense, and nobody has ever disclosed how the $110 million was divided.

There is a criminal coda. On 30 December 2020 Ticketmaster paid a $10 million criminal fine under a deferred prosecution agreement resolving five counts including conspiracy to commit computer intrusions, after employees repeatedly used credentials a former employee of a competitor had retained to get into that competitor’s password-protected systems. The Justice Department’s own words: “Ticketmaster’s employees brazenly held a division-wide ‘summit’ at which the stolen passwords were used to access the victim company’s computers.” The department never names the victim, referring throughout to “the victim company”; press reporting identifies it as CrowdSurge, the company Songkick merged with.

Hogarth’s own second act is stranger still. In June 2023 the British government appointed him to chair its Foundation Model Taskforce, reporting directly to the Prime Minister and the Technology Secretary — the body that became the UK’s AI Safety Institute. The announcement describes him as “the founder of the start-up Songkick and the venture capital fund Plural.” The member of this batch with the most public power holds it in artificial-intelligence policy, which has nothing to do with anything anyone built in 2007. His co-founder Michelle You now runs Supercritical, a carbon-removal marketplace that raised $13 million in June 2023.

What people get wrong

That the published batch list is the batch. Y Combinator’s directory returns nineteen companies for Summer 2007 and is missing two that contemporaneous reporting places in the batch by name and by position. It also lists one founder for a company whose acquisition coverage names three different people, omits a founder its own prose mentions, and files an acquired company as merely inactive. It is the best source available and it is not ground truth.

That the famous ones are from here. Justin.tv, Weebly and Octopart were Winter 2007. Heroku and Tipjoy were Winter 2008. Wufoo was Winter 2006. The Cloudant that IBM bought was Summer 2008, and the similarly named company at this Demo Day was a bandwidth router with no demonstrated connection to it. Only Dropbox and Disqus are genuinely Summer 2007 — and the Dropbox story everyone tells, Boston passing and Sequoia funding after the Mountain View pitch, is itself an argument that the room mattered less than which room.

That an acquisition means money. Six of the seven acquisitions here were acqui-hires or small asset sales, and the one that happened fastest — Anywhere.FM, inside months — was swallowed by a company that was itself sold for under a million dollars and shut down two years later. “Acquired” in a batch list is a status, not a sum.

That Dropbox’s falling market value means its shareholders lost money. It does not. The stock is above both its offer price and its first-day close; the market capitalisation fell because Dropbox bought back nearly half its own shares, which is capital returned rather than destroyed. The company got smaller on purpose.

That the failures were the failures. The second- and third-largest fortunes to come out of this batch belong to the founders of Meetyu, which Y Combinator records as having zero employees, and ContestMachine, which it records as no longer in business. Neither fortune has anything to do with the company they pitched. At least three of the batch’s founders ended up on the investing side of the table instead. If this cohort has a lesson about where money comes from, it is about people rather than companies — and a scoreboard organised by company cannot show it.

That the typical outcome here was a modest acquisition. It was not. Ten of the nineteen companies — a clear majority — are inactive with no transaction of any kind, and of the roughly thirty-four founders Y Combinator names across these nineteen companies, I could find no verifiable public record of what happened to about a dozen of them after 2010. The characteristic outcome in this cohort is not a small exit. It is no further public trace.

Bottom line

Nineteen companies pitched on the same two days in August 2007, funded on the same terms — about $15,000 for about six per cent — by the same two people, in the same programme, in the same city. One of them is now a roughly $7 billion public company controlled outright by a founder worth $2.4 billion. Seven were acquired for sums that nobody, in nineteen years, has ever disclosed. One still runs a browser game. Ten are dead domains. More than half the room is a zero.

This was supposed to be the series’ clean case: the cohort where the denominator is printed in advance, so you can finally calculate a real hit rate instead of reasoning backwards from the winners. And you nearly can. The honest rate for a verifiable nine-figure exit that reached the founders is one in nineteen, about five per cent — lower than the folklore implies, and roughly what any venture portfolio would predict.

But the number that matters most is the one the published list cannot produce. Eighteen of nineteen companies have no reliable figure attached to them at all, so the distribution of outcomes inside a knowable cohort is still unknowable. The second-biggest fortune from this batch belongs to a company the official record scores as nothing, built nineteen years later with no connection to it. And the list itself has lost two of its own members.

Which is the real finding. The accelerator model’s premise is that the batch is the unit — same room, same money, same filter, so the outcomes ought to be comparable. Summer 2007 is the best test of that premise anyone can run, because it was written down. And it says the batch is not the unit at all. The unit is the person, and the person takes nineteen years to resolve.

---

Methods and sources. The batch roster, company statuses, team sizes and founder lists are from Y Combinator’s own public company index, queried for companies whose batch field reads Summer 2007, which returns nineteen; individual company pages supply the statuses (one Public, one Active, six Acquired, eleven Inactive), the former-name records for Splashup (formerly Fauxto) and Hello Chair (which produced Adpinion), the ContestMachine description quoted in full, the Meetyu record (Inactive, team size 0) and its founder bio, the Blue Frog Gaming record (Active, team size 11), and Savraj Singh’s “YC S‘07 and W‘09” line. The same index supplies the batch assignments used to correct common misattributions: Justin.tv/Twitch, Weebly and Octopart in Winter 2007, Heroku and Tipjoy in Winter 2008, Wufoo in Winter 2006, Cloudant and Snipd in Summer 2008, Wattvision in Winter 2009. The nineteen companies are Dropbox, Disqus, Songkick, AppJet, ZumoDrive (Zecter), Anywhere.FM, Clickpass, SocialPicks, Blue Frog Gaming, Bountii, Biographicon, ContestMachine, Eivod, Fuzzwich, Hello Chair, iJigg/Top440, Meetyu, Reble and Splashup; this piece counts SocialPicks among the acquired rather than the inactive, against Y Combinator’s own status, because its acquisitions are documented. Demo Day quotes and the ten companies named that day are from TechCrunch, 16 August 2007, which describes the event as the “fall demo day” in the same sentence that gives the count of nineteen; the Versionate introduction is from TechCrunch, 11 July 2007. Versionate’s and the 2007 Cloudant’s absence from the directory was checked against the full nineteen-company list and against every populated former-name field; a third-party aggregator asserts Versionate and Zecter were the same company, and I found no support for that claim in any Y Combinator record, so it is not repeated here. What became of Versionate could not be established at all. Dropbox’s offer price, share counts, the 1-for-1.5 reverse split of 7 March 2018, the three Houston trusts and their dates, the two Ferdowsi trusts, the 2,333,333 shares each founder sold, the 180-day lock-up, the ten-votes-per-share Class B structure and the approximately 98.1% Class B voting power, Sequoia’s 87,171,450 Class B shares and Accel’s 18,752,565, and the Co-Founder Voting Agreement share blocks of 16,928,213 and 7,093,163 that account for the gap between what the founders owned and the percentages the prospectus reports, are all from the final prospectus of March 2018; the $1,780,032,534 and $719,685,078 figures are the trust-held share counts multiplied by the $21.00 offer price and are the value of shares owned, not of total beneficial ownership. The nine tranche prices, the single vested tranche of 2,066,667 shares on 15 November 2021, the 8,266,666 shares still outstanding, the 23 March 2028 end of the performance period, the $10.60 average grant-date fair value and the $109.6 million expense for Houston’s grant alone, Houston’s 74,616,889 Class B shares and his 98.99% of Class B and 82.40% of total voting power as of 31 March 2026, and the 2025 figures of $689.1 million of operating income at a 27.3% margin, are from the 2026 proxy statement. Ferdowsi’s resignation and quotation are from the Form 8-K of March 2020. Market capitalisation of about $7.1 billion at $32.70 a share on 29 September 2026, $11.18 billion at the 23 March 2018 close, and the fall in shares outstanding from roughly 392 million to about 217 million are from StockAnalysis; the $28.48 first-day close is from CNBC’s coverage of 23 March 2018; these move daily. Houston’s $2.4 billion is Forbes as of 26 September 2026 and Ferdowsi’s $400 million is Forbes as of December 2016, pre-IPO and not updated since. Y Combinator’s 2007 terms and the $15,000 average are quoted from TechCrunch, 16 March 2009, which is contemporaneous but secondary; Y Combinator’s own 2007 FAQ survives only in the Internet Archive and was not reachable for this piece, so the figures have not been confirmed in Y Combinator’s own 2007 words. The 2010 terms are from Forbes and the current deal from Y Combinator. The Livingston cheque date, the Boston rejection and the 16 August Mountain View presentation are from Y Combinator’s own post. Acquisition terms and the absence of disclosed prices are from the linked contemporaneous reports in each case, cross-checked for Disqus against Zeta Global’s 2021 Form S-1, where the Disqus acquisition-related liability runs $6.451 million at 1 January 2019 less $4.649 million paid and $1.802 million of gain on extinguishment to nil, and no purchase price appears; the DISQUS and LiveIntent data-collection language is from Zeta’s FY2025 Form 10-K filed February 2026. The imeem resale figure of under $1 million is a press report, not a disclosure. The Songkick sequence is from the linked TechCrunch reports of June 2015 and July 2017 and from Complete Music Update on the January 2018 settlement; the $110 million figure is reported as fact by multiple outlets and as “rumoured” by at least one, and no party has disclosed how it was allocated. The Ticketmaster fine, the five counts and the quoted language are from the Department of Justice release of 30 December 2020, which does not name the victim company. Hogarth’s appointment and the quoted description are from gov.uk, June 2023. LiveKit’s $100 million Series C at a $1 billion valuation is dated 22 January 2026; OpenAI is named as a customer on LiveKit’s own site, and Salesforce Ventures participated in the round. Rob Fitzpatrick’s words are from his Indie Hackers AMA, where the misery he describes spans several of his early companies rather than Fuzzwich alone. Domain states were observed on 1 October 2026 and are described in general terms because they are volatile; the Blue Frog Gaming figures are live counters on the game’s own page and were rounded for the same reason. Founder counts are mine, from Y Combinator’s nineteen company pages, and total thirty-four; the statement that about a dozen founders have no verifiable public record after 2010 is the result of my own searching and is a statement about the absence of findable evidence rather than about what those people did. No source discloses what any Summer 2007 founder other than Houston and Ferdowsi realised, and this piece estimates none. This draft was fact-checked line by line before publication and the check changed several substantive things: an earlier version reported Sequoia’s stake using a pre-split share count and presented Sequoia’s and Accel’s share of Class B as their share of the company; it concluded from Dropbox’s falling market capitalisation that shareholders had lost money, which the buyback history contradicts and which is now a corrected entry in “What people get wrong”; it found one hole in the published roster where there are two; it attributed the founders’ disclosed ownership percentages to their trust-held shares, conflating ownership with voting control; it said seventeen of nineteen companies lacked a reliable figure where the correct count is eighteen; it built the median-outcome argument on founders rather than companies; it listed Tesla as a LiveKit customer, which LiveKit’s own customer page does not support; it described Rob Fitzpatrick as angry, which his own account does not say; and it carried an unsourced remark comparing Clickpass’s price to Digg’s, which has been removed.

Related reading: The Class Of: The PayPal Mafia · The Class Of: The WhatsApp 55 · The Class Of: The Google Class of 2004 · Venture Capital: The Culture of Tech Money · Equity Compensation: RSUs, ISOs, and the Tech Wealth Engine · Paths to Millions: How First-Generation Wealth Is Actually Built · Falls From Grace: Bankruptcies, Frauds, and Reversed Fortunes