Summary

One deal can save a $12B fund — but only via bet sizing. Tiger’s 350-deal 2021 vintage looks salvageable because of OpenAI and Scale, and Rory O’Driscoll’s math is blunt: if they did one 350th in OpenAI, even a 30X doesn’t matter; if they had “the guts and the foresight” to put 10-15% of the fund into something that 7-8Xs, they’ve “snatched victory from the jaws of defeat.” His line for the episode: “It’s just hard to dig out of a $12 billion hole in $100 million increments.” Jason’s contrarian add: Tiger is unfairly criticized — the back catalog will 1X on liq prefs and “there’s actually quite a few that are really f*ing good.”
Perplexity’s rumored $500M at $14B (Accel-led) is priced as a lottery ticket, and Rory thinks that’s rational: “a credible one in three, not equally weighted… OpenAI’s clearly gonna win, but maybe you can be third, and that’s worth a damn sight if the prize is a trillion bucks.” Jason Lemkin’s pushback: they announced $100M ARR, call it a generous $200M now — is this “a distant number three getting a massive premium?” Jason’s half-joking memo reform: the first row of every IM should be “odds of trillion-dollar outcome” — “if it’s north of 2%… we do the deal.”
Coatue’s $50K-minimum retail fund will sell — the question is whether it should. Rory: “It is the only way you can get access to those assets” — the risks aren’t manager quality but cycle timing and redemption gates, and the zoom-out is “madness… these companies should be public.” Jason smells Boiler Room: he tells every individual not to LP in venture funds, including his own, because nobody understands illiquidity — “VTI is the perfect product for 99% of people.”
The Klarna “walkback” is misread — the AI slider only moves one way. Jason’s ground truth from an app he calls Gorgeous’s 20,000 SMB support customers: average AI automation is 20%, engaged users hit 40%, and exactly two out of 20,000 went to 100%. Klarna moved the slider to 100, went too far, and is rehiring maybe 200 — “800 will still be AI’d.” Rory closes the book on their bet (“You’ve been margin called”) but agrees on substance: LLMs take support to 50-70% with improved NPS — “this is just a five-year trend, not a one-year moment.” Jason’s faster clock: every tech job changes “by the middle of next year,” and Microsoft’s 3% layoff is “not enough.”
SBF as the era’s best investor is only half a joke: FTX’s early Anthropic and Cursor positions alone would have repaired its financial performance. Rory: “he had an excellent business, blew it with fraud, and did some great venture. That’s quite a polymath.” On a second act they split — Jason thinks someone gives him “$100 at a billion post” out of a $5B fund; Rory says the convicted-criminal bar is categorically higher than WeWork-style implosion: “once you’re past the convicted criminal stage, the bar goes way up.”
OpenAI now has two non-technical CEOs — and Jason thinks that’s genuinely weird. With Fidji Simo as CEO of Apps under Sam, “one of the greatest technological innovations of our lifetimes is now run by two non-technical people,” and in his experience non-technical CEOs “almost all fail.” Rory shrugs — “we are 10 years into the least normal startup on the planet, why stop now?” — and notes the PBC restructuring lands OpenAI exactly where Anthropic started: “Anthropic did it from day one. They got it right.” Unwinding Microsoft’s 10%-of-revenue-until-AGI deal, plus Elon litigation, means legal bills Jason guesses at $10M a month.
Rory’s portfolio math has repriced: fewer, bigger winners means more at-bats. Rory’s target portfolio has moved from 20 deals to 25-28, because exits now compound from $150M-type outcomes to $300M+ over two-three extra years in which “one in three of these companies will f* up.” OpenAI (four last year, “12-ish” this year, with another figure trailing off, and leaked forecasts growing faster than Google) could hit today’s-Facebook valuation before ever going public — but he won’t concentrate blindly: the Nifty Fifty took 10-15 years to come back, the 2000 Nasdaq took 14.
Clay at $1.5B is the Hopin of AI fear — and that’s a compliment with a warning. Jason: every scared 2021 CMO is “hiring Clay consultants… writing checks like Hopin in 2020,” budget appears overnight when six figures might save your job — but “the knives are out”: a 19- or 20-year-old Stanford dropout competitor is already at $2M in a couple of months. His advice: “I would armor up if I were Clay… raise another 100 million and just scorch-earth everyone in the space.” Rory’s key distinction from Hopin: selling fear ahead of the product is legit “provided the product gets there in the end” — Hopin’s end demand died; this one’s ground is real.

Deep dive

1. Owner’s $120M at $1B — “in the zone, there’s unlimited effing capital”

Jason led Owner’s seed and plays at not reading the docs — “all I care about is my ownership” — before confirming the numbers: $120M at a $1B valuation, on roughly $40M growing 10% a month. His benchmark check: “triple, triple, double, double’s still good enough,” and Owner is objectively faster.
The process was the classic 2021 shape that still gives him hives: Adam socialized with VCs for months, then “open the data room on Monday, get two term sheets that afternoon, and get all of the term sheets by Wednesday.” Insiders alone could have filled $150M. The meta-lesson: “If you’re in the zone… there’s, like, frigging unlimited capital. But you gotta be in that box” — and the box is harder than ever to be in.

2. Hearing about a deal Monday with a Wednesday deadline means you’re already behind

Harry’s team keeps hitting “great founder, decide by Wednesday” on a Monday. Rory’s answer: the framing of the question is the failure — someone else “met them before in the last round… has done some work the prior Monday to Friday. For them, Monday is just confirmatory.” You can assimilate data in two days; you cannot assimilate the person.
His operating discipline: a Salesforce hot list of 10-20 names he could imagine backing in the next 12 months, or he’s not doing his job — “you can woo 10 people, you can’t woo 200.”
On preempting a month after the seed: Rory has never paid twice the three-months-ago price — “that feels like an idiot” — but he’s wrestling with whether the bias is wrong, because the best companies raise with the highest velocity. His Adam Smith framing of OpenAI: “We put in 100 million, the model got smart. Shit, let’s do a billion. Oh, got smarter. Let’s do 10.” Jason’s shorthand for Sequoia’s serial re-entries (seed, then Clay at $1.5B): “It’s win when you can win.”

3. Series A down 81% — and for the founder it’s 0 or 1, not a statistic

Carta says Series As are down 81%. Rory’s mechanism: seed is the believe-in-the-team round, the A is show-me-the-traction round — “belief is easy to manufacture and traction is hard.” Smart founders who lack it aren’t raising at all: like a barrister, “never call a question unless you know the answer.”
Jason’s zero sympathy: “Cry me a river… Be the best of breed, you’ll get five term sheets.” He lived sequential years of “easy to get funded, impossible” as a founder; complaining is “a B tier way to look at it.”
Rory’s startling analogy, from life: he had stage-four colon cancer just under 40, and internalized that survival statistics are for doctors — “to the patient, it’s 0 or 1. You either make it or you don’t.” Same for a startup: either you have something worth funding and the statistics don’t matter, or you don’t. Asked why he came back to venture at all, no yurt, no Everest: “I like my work and I just like to keep doing it, and you just do it until you can’t.”
His “yay venture” coda: he was among the first patients on Avastin — Genentech, Kleiner-funded — and received it literally a week after it was FDA-approved. “Before that you’d have been toast.”

4. Tiger: one deal can save a $12B fund, but only if the bet was sized to matter

The setup: ~350 deals in 2021 (deals, not necessarily As — “don’t quote me”), a $12B fund, and now OpenAI and Scale positions that could rescue performance. Rory refuses the morality play: “The facts come in… the numbers tell.” The only variable that matters is bet sizing — one 350th of the fund in OpenAI bails out nothing even at 20-30X; 10-15% in something that can 7-8X and “maybe they’ve snatched victory from the jaws of defeat, and more power to ’em.”
Don’t conflate vintages: any 2021 exit is a 2018 late-stage deal, so the definitive late-stage firm “probably has a great ‘18 fund” — buying at $2B in 2018, selling at $6B in 2021. Then “you start thinking you’re smart ‘cause you’re making money, and then you do 315 more deals in 2021.”
Jason breaks ranks: Tiger is unfairly criticized — if LPs hold through OpenAI and Scale, the back catalog will 1X on liq prefs, “they won’t lose money, and there’s actually quite a few that are really f*ing good.” Rory’s memorable frame for the pivot-to-concentration: “It’s a little like Bush in Iraq… a mistake to go in, but at least in 2006 he surged his way out. This could be the Tiger surge.”

5. SBF picked Anthropic and Cursor early — “that’s quite a polymath”

Jason’s observation: FTX’s positions in Anthropic and Cursor alone would have saved its financial performance, given time and no commingling. Rory is genuinely impressed: picking two of the most important post-crash companies in ‘21/early-‘22, pre-ChatGPT moment, “is astonishing. What a talent.” And the core exchange “worked fine, too… he had an excellent business, blew it with fraud, and did some great venture.”
Jason’s deadpan riff — flagged by Harry, for the record, as irony not thumbnail material: “I tell founders, just only do, like, a little fraud… he just did it too far. You poke the bear too much.” And the kicker: “I think all of crypto’s a little fraud.”
On a second act they genuinely split. Jason: “I think someone will fund him… just, like, 100 to start, at a billion post… It’s 2% of a new $5 billion fund” — Theranos and the Synapse guy got theirs. Rory draws a hard line between dodgy-aggressive failure (WeWork: “hubris, grandiosity, implosion… but no convicted crime”) and fraud convictions: “once you’re past the convicted criminal stage, the bar goes way up.”

6. Coatue’s $50K retail fund: democratization, or Boiler Room?

Jason’s gut, watching Philippe Laffont market the fund on All-In: “Vin Diesel. Let’s just rip off the retail investor… I’m not a cynic, I think this is cynical.” The fee structure — 12.5% carry and 1.6842% fees — is “too insider baseball-y” for retail to even parse. Rory concedes the blend is fair math (half public assets at index-like cost, half venture at two-and-20 equivalent): “It’s not cheap. It’s market.”
Jason’s structural read: every big PE firm is making the same move toward private wealth, “anything that works in finance gets copied immediately,” and there’s surely “a PowerPoint on the desk at Andreessen, at General Catalyst, and at Lightspeed.” It will sell because the pitch is true: “it is the only way you can get access to those assets.” One puzzle he can’t square: Michael Dell and Bezos as anchor tenants — “they pass the 50K threshold, let’s put it that way.”
The real risks aren’t manager quality (Coatue is “not Benchmark or Kleiner” but in the flow, with access): they’re cycle timing — “if the next five years are tough, no one can save you” — and redemption gates, per the Blackstone/BlackRock real-estate fund that had to gate liquidity. Jason’s harder line: he tells everyone not to LP into venture funds, including his own — an 8X that pays 17 years later “is not worth it for the stress. VTI is the perfect product for 99% of people.”
Jason’s zoom-out, the sharpest sentence of the segment: “This is all madness. These companies should be public, and then Fidelity Growth Fund could do them at 50, 70 bips… This is a workaround for a problem that would be better solved some other way.” Harry’s needle: a $1.2B fund “doth butter no parsnips” at that scale; Rory: “the way you get 10 billion is you start with one billion.”

7. The Klarna walkback is a slider correction, not a reversal

Jason, defending his 100K bet that AI replaces jobs fast, brings actual data: at an app he calls Gorgeous, 20,000 SMBs run an AI-support slider from 0 to 100. The average is 20%, engaged customers reach 40, and exactly two out of 20,000 went to 100. Klarna’s Seb did what those two did — slid to 100, overshot, and is walking partially back: “He’s probably gonna rehire 200, and 800 will still be AI’d… every three months that slider’s gonna move closer to 100.”
Rory jokes the bet is over (“You’ve been margin called. Send the money”) but is in “violent agreement” on mechanism: pre-LLM you chipped away 20-25% of tickets — the easy ones — while LLMs reach 50-70% with no service deterioration and, per his references, improved NPS. His hedge against Jason’s speed: “This is just a five-year trend, not a one-year moment.”
Harry’s strategic read on the timing: Klarna sang AI-first when it needed an IPO story — his Duolingo guest was explicit that “public markets take a very binary approach: you’re either an AI winner or an AI loser” — and now that Klarna isn’t going public, Seb can bank the honesty for the next window.
Jason’s darker gloss: these statements are shock therapy for the org. “A lot of big company CEOs I talk to are honestly like, ‘I’m not sure I need 80% of my team today’… Microsoft laid off 3% of their company today. It’s not enough.” His timeline, categorical: “by the middle of next year, in tech almost every single person’s job will have changed.” His one urgency idea for a $100M-revenue CEO: force full RTO in 30 days and let everyone but S-tier engineers go — while admitting he himself only goes in two days a week: “I acknowledge it’s hypocritical.”

8. Two non-technical CEOs now run the most important technical company alive

On Fidji Simo as OpenAI’s CEO of Apps, Jason plants a flag he keeps returning to: “I just think it’s really weird… one of the greatest technological innovations of our lifetimes is now run by two non-technical people.” His pattern from watching outside CEOs take over his investments: “they never understand the product… they almost all fail” — even granting Sam is “S-S-tier” at recruiting and she’s S-tier at operating.
Rory’s counter is half a shrug, half awe: “They weren’t gonna start being normal now, dude. We are 10 years into the least normal startup on the planet. Why stop now?” — and the fact the most compelling technical product in 15 years was built under a non-technical CEO “speaks to some shrewdness by him… a stunning achievement.”
On what Apps means: Rory expects broadly horizontal plays — chat, coding, consumer, and especially shopping — not customer support (“too idiosyncratic”). His consumer maximalism: “ChatGPT’s gonna suck in all your brain… Remember when you thought Evernote would know everything about you? You ain’t seen nothing yet.”

9. Fewer, bigger winners — so buy more at-bats, but don’t become the Nifty Fifty

Harry’s dare — why not just chuck the fund into OpenAI for an easy 3X? — gets Jason’s David Sacks story: Craft fund one reportedly put a third of the fund into SpaceX in week one (“I might have this a little wrong, but I think it’s mostly correct”), looked nutty, “obviously worked.”
Rory’s repricing of his own model is the analytical core: Rory’s target portfolio has moved from 20 deals to 25 to 27, 28 — because exits now run to $300M-scale outcomes over two-three extra years in which “one in three of these companies will f* up.” Fewer winners, but compounded bigger. Taken to the limit, OpenAI “might get to Facebook-type valuation before it ever goes public. Facebook today, not Facebook when it went public.”
His restraint isn’t squeamishness, it’s history: “You can take a good idea and push it to such extremes that you end up wrong” — the Nifty Fifty in ‘68 didn’t come back for 10-15 years, the 2000 Nasdaq took 14. On OpenAI’s numbers as he sketches them — four last year, “12-ish” this year, then a trailing-off figure, with leaked forecasts above Google’s equivalent trajectory — “at some point you’re over-extrapolating into the future, and when is that point? I don’t know.”
The practical bar to entry anyway: OpenAI’s last round was “show up with 250 million or don’t show up at all.” Rory, with his $30M check: “I can make a phone call, but I don’t think they’ll get back to me… Jason’s AI will look at my ticket and say ’not worth replying to.’”

10. OpenAI’s restructure lands exactly where Anthropic started

The resolution isn’t a return to nonprofit purity: the operating company becomes a public benefit corporation (Patagonia-style), with the nonprofit foundation keeping board control and a significant economic stake. Rory’s approval is backhanded but real: “That kind of structure is actually what Anthropic did from day one. They got it right. OpenAI is basically saying we’re gonna go for the structure Anthropic already has… somehow they get there, but God knows how.”
The Microsoft angle flipped in hindsight, on Jason’s telling: what looked like “a crazy, weird get-around-M&A-antitrust game” where Microsoft would “lose billions subsidizing a backup bet” now looks like “an epic investment” — everything they got plus 10% of all revenue until there’s AGI. Which is precisely why unwinding is brutal: converting a profits-first-then-rev-share-then-cap claim into a simple ownership percentage when “everyone has a veto” and Microsoft’s leverage is strong.
Add Elon’s litigation risk over whether even this is not-for-profit enough, and Rory’s forecast is for the lawyers: “a lot of kids are gonna get put through college on this deal.” Jason’s estimate: legal bills of maybe $10 million a month.

11. Perplexity at $14B: paying up for a one-in-three at a trillion-dollar prize

The rumor: $500M at $14B, led by Accel. Jason likes the product (“the closest to Google on steroids”) but questions the price: they announced $100M ARR recently, so “let’s say they’re at 200 today, which is still generous… is this a distant number three getting a massive premium?”
Rory’s framing carries the segment: only three players are relevant — OpenAI, Anthropic, Perplexity — and Perplexity was “first with the idea of web search plus LLM combined. That gives you the right to play the game.” The pitch: “an at-bat, a credible one in three, not equally weighted… OpenAI’s clearly gonna win, but maybe you can be third, and that’s worth a damn sight if the prize is a trillion bucks.” His ‘96 precedent: four search engines went public that year — Lycos, Excite, Yahoo, one he can’t recall — because a big obvious market gives everyone a chance. Plan B: trillion-dollar acquirers “who are like, ‘Shit, maybe I’d like to mess with Google’s head.’”
Harry discloses he’s a Perplexity investor and adds the under-covered edge: distribution through big European telecom partnerships — default placement on consumers’ phones, exactly Google’s early playbook.
Jason’s memo reform, only half facetious: his first-ever IM (Pipedrive, 2013) scored “5% chance of a $100 million exit” — it sold for $1.25B. The new row should be “odds of trillion-dollar outcome,” and “if it’s north of 2%, we do the deal.” Rory’s inversion of the same idea: “We’re all upside junkies… if a deal has all the risks of a classic private company but no embedded upside beyond the base case, you probably shouldn’t do it. This one has it in spades — it’s trillion-dollar pixie dust, not the $100 billion kind.”

12. Clay is Hopin-for-marketers — sell the fear, but the product has to land

Against Harry’s thesis that Clay’s $1.5B Sequoia secondary prices it as “the next credible threat to Salesforce,” Jason reports from the field: alongside genuinely sophisticated teams, “every 2021 CMO that is scared they’re gonna lose their job… is hiring Clay consultants, Clay engineers, writing checks like Hopin in 2020.” Budget materializes instantly when six figures might save a career: “My budget’s five, ten million in marketing — all I have to do is come up with 200K for Clay? Send over the contract tonight.”
The warning inside the compliment: “the knives are out.” The night before, a 19- or 20-year-old who dropped out of Stanford with Jason’s daughter told him his Clay competitor hit $2M in a couple of months (“we’re just much easier to use and we have better data sources”). Gong took four years for competitors to decode; “now it’s happening in days or weeks.” His prescription: “I would armor up if I were Clay… forget the secondary, keep it as primary… raise another 100 million and just scorch-earth everyone in the space.”
Rory grounds what Clay actually is: a pre-AI RevOps product that brilliantly consolidates four or five data sources into one clean 10,000-name list, now layering Clay agents down-funnel toward AI SDR — “so far from Salesforce CRM that I can’t even think about it in any useful fashion. Which should never get in the way of a good story from a VC.” On selling fear ahead of the product, his verdict is conditional: “If you seize ground but the ground turns out to be waterlogged, you’re toast” (Hopin’s demand died); if the ground matters and the product catches up — as he’s watched happen in white-collar professional services — “then you win. It’s a totally legit tactic at this point in the cycle, provided the product gets there in the end.”
The closing spread on pace: Jason says being the vendor to take advantage of AI fear is a great strategy; Rory says lean in or perish (“If you’re resisting it, you’re toast”) but sees the automation scale advancing “5 or 10% a year”; Jason’s correction, flat and immediate: “A month. Five or 10% a month.”

Verification Notes

The raw captions render the company name as “Gorgeous”; the digest avoids asserting the inferred name “Gorgias.”
The raw captions say “Scale,” “Cursor,” “Opus,” and “PFP/PSP”; familiarized transcript forms were restored.

AI is Making Enterprise Search Relevant, with Arvind Jain of Glean Google AlphaEvolve - Discovering new science (exclusive interview)