a16z's $20BN Fund & Founders Fund's $4.6BN & Why Josh Kushner Has Mastered the Game

Summary

• Spreadsheet SaaS investing is contested, not simply dead: markets saturated (“anyone who needed a Zoom or DocuSign account got one in code”) while AI rewrites the game every six months. The killer line — “it used to take you five years to fall out of product-market fit. Now it can be five weeks” — means investors are underwriting upside while knowing less: “you’re just taking on more risk for a dollar of revenue,” and if you also overpay, “it’s a sucker bet.”
• The $3 trillion overhang is the industry’s real problem: roughly $2T of mature, slower-growth private SaaS has no IPO trajectory, and PE isn’t tire-kicking because venture funded exactly what PE hates — horizontal markets with no pricing power, where cutting R&D loses “30-40% of your revenue” in two years. The path out is “grim industrial work”: private-to-private mergers, grinding to profitability, small IPOs.
• Mega-fund logic pencils — until multiples compress. Rich’s defense of Andre’s $20B: if you see every S-tier deal, the counterfactual spreadsheet (Databricks at $27B in 2021 would already be 2.5x) says you can deploy it in 24 months. Rich’s pushback: seeing every deal means seeing every bad deal — “there are 99 [bad] deals for every one good deal” — and the game only stops when LPs’ CIOs stop allocating.
• The Thrive doctrine wins this cycle: “buy the best damn house on every block” — Stripe, OpenAI, Databricks — “then you just go home and wait for the checks to roll in.” Rich’s now-stolen corollary: when outcomes run north of 2,000 billion, “seed is for suckers.” The buried risk is correlated multiple compression: if growth P/Es go from 30-35 to 9-11, with growth stocks at 12-13, a marquee-asset strategy built for 3x “could get a 0.5 or 7x.” “There’s no such thing as free money.”
• Founders Fund is the vindicated exemplar — $4.6B raised, $1.6B oversubscribed, and a leaked track record showing exactly what they advertise: 15-year holds (SpaceX since ~‘07-08) compounding 30-40% into “an 8 or 10x fund,” plus 30% fund concentration. Rich to LPs trying to clone it: “Ten other funds won’t be just like Founders Fund, because they’re not the same people.”
• Staying private is a “public policy failure”: the same Stripe compounding 15% gross nets savers 14.3% via Fidelity at 70bps but ~10% via venture at 2-and-20 — “a monstrously stupid outcome” that ends only when late-stage privates underperform publics by the fee load and capital reallocates.
• SSI at $32B is rational — fund the people with the “secret recipe” (OpenAI-alumni labs worked; others didn’t, “with the exception of Grok”) — and “there is zero chance this does not get bought for at least the pref. It’s [expletive] Ilia.” But don’t bank the pref: acquirers now “don’t give a rat’s ass what the certificate of incorporation says” and route acqui-hire dollars around VCs.
• When the tide goes out, GAAP is what’s left. Galbraith’s “bezel” — boom-time embezzlement surfacing in the bust — frames Deel/Rippling, and Jason’s survey of 2,000 B2B sellers found 93% lie to win deals. Today’s ARR is “neither A nor R — and may or may not be revenue.” Parting split: OpenAI at $300B — guests out “north of 100,” Harry all-in (“escape velocity reached”); Cursor at $10B hinges entirely on whether IDE revenue is durable.

Deep dive

1. Spreadsheet SaaS investing is contested — the 20-year playbook ran out

• Harry opens with Victor Lazerte’s (Benchmark) claim that spreadsheet SaaS investing is dead. Rory (the Box/DocuSign-era investor) agrees on the mechanism: for 20 years the direction was obvious — “take X, move it to the cloud, compound” — so all that was left was math, “evaluate the relative growth rates and pick the most efficient.” He invested in Box in 2010 and it didn’t change through 2024.
• Jason (he cites his SaaS survey and the EchoSign-vs-DocuSign rivalry) recalls the old yardstick Rory gave him: “1 to 10 in five quarters or less is S tier.” By late 2020 it seemed like every startup met it — one top cloud VC offered two of his portfolio companies high-nine-figure term sheets “without talking to the founders.”
• What killed it: two things at once. Existing markets saturated — “anyone who needed a Zoom account or a DocuSign account got one in code” — while AI startups took off in a regime where “this [stuff] changes every six months.” Rory: “I’ve had companies acquire and lose product-market fit two or three times in a two-year period.”

2. Product-market fit now decays in five weeks, not five years

• The episode’s sharpest line: “It used to take you five years to fall out of product-market fit. Now it can be five weeks — that’s not just a catch line.” When all three started, PMF plus a decent team bought five years of runway.
• Rory’s two causes: model progress at a deep level, plus “we’re still at the figuring-out stage” — even absent better models, customers change how they want it within months. His analogy: 1999-2003, when “what a SaaS company was” was unsettled before Salesforce nailed it. Harder now because “instead of automating some back-office [stuff], you’re really trying to automate the head of the worker.”
• The intensity shift is real: the best startups are “in the office, seven days a week, 12 hours a day” versus 2021 competition “working 10 hours a week from home — for real, not for fake.” Jason: “If you haven’t evolved, you’re going to die.”

3. You’re underwriting upside while knowing less — expect bimodal fund results

• Harry’s core question: with transient PMF and sugar-high revenue ($20-50M scaled fast), what exactly are we underwriting? Jason: “What you’re underwriting is the upside… you know less at every stage on every check you’re writing today than 10 years ago. You’re just taking on more risk for a dollar of revenue.”
• Harry’s pushback — prices are inflated, so you’re not being paid for the risk. Rory concedes the trap: upside arguably bigger than SaaS, but pay up so much that “even that upside has been competed away — then it’s a sucker bet.” “Every time you’re writing a check today you’re going: I know a lot less… and paying a bit more.”
• Will fund results be bimodal? Yes, for two compounding reasons: more per-deal risk and elongated holding periods — roll the dice three more times and winners and losers spread by definition. “Portfolio construction really matters here.”
• Against Benchmark’s “portfolio construction doesn’t matter” (8% of fund, $55M first check into HeyGen; “they can raise at will”), Rory: it always matters — brand names hit bumps too. He raised his first independent fund in 2009, “in the Lehman and AIG offices the day they went bankrupt”; it took a year. “Probably true for Benchmark and KP — probably not true for the other 898 funds.”

4. Triple-triple-double-double SaaS can still raise — the problem is the middle

• Jason would fund non-AI SaaS growing 3x/3x/2x/2x “all day every day — if you’ve got one of those, call me” (he passed on exactly one, “and I’m an idiot”). Rory: that’s not the issue — the real problem is the myriad companies at $50M growing 10-20%, or $100M growing 8-9%.
• But behavior has shifted: Jason estimates 70-80% of the SaaS investors he grew up with won’t take those meetings — “they’re momentum investors; they want to put 200 million into the latest AI deal and triple it in eight months.” Rory’s steelman of the heuristic: “I’m not going to kiss all those SaaS frogs… anything that could have been done 20 years ago in SaaS probably has been done.”

5. The $3 trillion question — and PE isn’t riding to the rescue

• Rory sizes the overhang: ~$3T fair value of private venture assets, of which maybe $0.5-1T is high-growth new stuff and “the other two trillion is mature, slower-growth SaaS and cloud companies that don’t have the trajectory anymore for an IPO.” Unlike 1999-2002, “you can move on from $200 million… you can’t walk away from $2 trillion.” Expect “a huge amount of really grim industrial work”: grind to profitability, private-to-private mergers, small IPOs — “case-specific, long and tiring.”
• Jason’s alarm: PE isn’t even tire-kicking anymore — he used to hear “20 firms called” at every SaaStr Annual. Rory’s why: PE loves “a boring-ass software company in a teeny-tiny vertical with 40% market share where they can screw the customers for the next five years by raising prices.” Venture funded broad horizontal markets — fail to get the billion-dollar outcome and you’re subscale with no pricing power, “and PE guys just hate that.”
• The durability trap: in a 2016-era horizontal SaaS company, cut R&D and sales and marketing and “your gross dollar retention will be 80%”; you won’t sell anything new, you’ll decline, and the product becomes irrelevant in two years. Harry’s last PE exit (~$300M legal-tech deal): sales and engineering gone, revenue durable “at least for a while”; in two years, “you’ve lost 30-40% of your revenue.” Coupa, Anaplan, Zendesk (likely) had scale and are “probably growing in the teens” — below $400M revenue it gets much harder.

6. The $20B fund math: seeing every deal isn’t the problem — picking is

• On Andre’s $20B fund and General Catalyst’s $8B: they’re obviously playing for a small number of $10B-100B outcomes, not billion-dollar exits (“thanks for paying for the Christmas party” — at 8% ownership a $1B exit returns $80M). Rich: “No one gets given $20 billion because they’re idiots.”
• Jason’s spreadsheet defense: firms like that see every deal, so run the counterfactual — Databricks at $27B in 2021; $3B in would already be 2.5x. Add it up “and it maybe solves to 20 billion… they can deploy it in 24 months.”
• Rich’s pushback: “Of all the things that could go wrong with a $20 billion fund strategy, not seeing every deal is not a top-three issue.” See every good deal and you see every bad deal — “there are 99 [bad] deals for every one good deal” — so picking still matters; the binding question is whether privates can absorb the money. It stops only when “the LPs’ bosses, the CIOs, stop allocating” — and if that happens the day after you close, “you win the best, because you have 20 billion and no one else has any.”
• Absorption looks fine for now — OpenAI raising $30B, Anthropic’s multi-billion round, and the 2022 crash caused “no more than a pause for breath.” Harry’s sharper question: is it still a venture game? Investors who entered a model provider at $4B are only 3.5x up at $60B because employee dilution and stacked rounds ate the multiple.

7. The Thrive doctrine: buy the best house on every block — “seed is for suckers”

• Rich’s admiring anatomy: “A real estate investor knows only one thing — buy the best damn house on every block.” Fintech block: Stripe (likely — “Strike”), tick. OpenAI block, tick. Infrastructure: Databricks, tick. “Then you just go home and wait for the checks to roll in… I’m profoundly jealous of that insight.” The only criterion: “Is it going to make you money — and across the cycle? The first part looks like yes. The second: call me in a year or 10.”
• Rich’s corollary, which Rory vows to steal without credit: “Why struggle to pretend you can do 8x over 20 years on a seed fund when you can write one big check into a winner? The multiple will be lower, but the absolute return will be higher… When outcomes are a billion, seed is great. When outcomes are north of 2,000 billion, seed is for suckers.”
• The buried risk is correlated: “You’ve bought the best assets; the only risk is that the world decides equity isn’t worth as much.” If high-tech P/Es fall from 30-35 to 9-11, while growth stocks fall to 12-13 — his Nifty Fifty 1968-82 reference — a marquee-asset strategy built for 3x “could get a 0.5 or 7x,” while Rory’s lower-entry approach degrades to 1.5x.
• And the compounding assumption can fail: five or six $1T companies prove OpenAI can go from $300B to $1T — “what you forget is most tech companies don’t.” “Be a bit of a bummer to discover you’d invested in BlackBerry… There’s no such thing as free money — when stuff looks like free money, it typically means the risk isn’t fully recognized.”

8. Fund size is the strategy — and survival beats spectacle

• Why doesn’t Rory raise much more, given his track record? “I’m branded by surviving 99 to 2010… I don’t want to take a lot of money at the end of my career, manage it badly, and fail.” He’s prouder of small profits from 2000-2010 than bigger returns since — 70% of the people he knew in 1999-2000 were out of the business four years later.
• The casualty list makes the point: the two biggest momentum players of the decade, Tiger and SoftBank, are “already out of the game”; Insight survived by doing deals at earlier, lower prices. And even likely Wiz humbles the math — a reported $2.6B return in an $8.5B fund is a third of the fund. Jason: “I would probably quit venture if I did Wiz and it was only a third of the fund.”
• Some fund inflation is just arithmetic: nominal GDP is 3x since 1999 ($10T→$30T), so a $100M fund then needs to be $300-400M now; Emergence’s new $1B fund reflects roughly $50-60M checks (e.g., into Bolt) that used to be $15M. “You’ve got to size the fund for the strategy, because fund size is the strategy.”
• Rory’s construction discipline, contra Benchmark: given PMF variance he wants closer to 25 deals per fund (implying $700-800M at A/B check sizes). Harry’s pet peeve cuts the same way: $50M seed funds are $40M investable against $3-5M rounds — forced concentration risk.

9. Founders Fund: the leaked track record vindicates concentration and 15-year holds

• Founders Fund raised $4.6B with $1.6B oversubscribed — Harry has “never had such institutional demand for any single fund asset.” Rich: no surprise — “they may be, to a rounding error, the best fund,” and “my Bayesian prior on financial matters in venture should be checking what Peter Thiel likely does.”
• The leak showed exactly what they advertise: 10-15-year holds (SpaceX bought ~‘07-08) — compound a winner at 30-40% for 15 years instead of eight and “you end up with an 8 or 10x fund” — plus massive concentration, enabled by not caring if LPs dislike the risk. Brian Singerman once told Harry: “The enemy of great venture returns is capital concentration limits — we have 30% of a fund in certain assets.”
• Rich’s both-sides: concentration is the enemy of greatness and “the protector of massive wipeouts” — they had “the stones” to put $300M into biotech Stemcentrx (likely), took a 5x off the table, and three years later the acquirer cancelled the program. “Never forget the risk was there.”
• On the explicit no-B2B stance: they want “singularity deals” — N-of-one, high technological component, low competition — and almost nothing in B2B has SpaceX’s uncontested space; the counter is “there have been two to three hundred SaaS winners.” Jason’s inside detail (via a friend who worked there, likely Sam Blond): they classified Ramp as a fintech. And the ~50% GP commit shapes it — “If I had that much, I wouldn’t want to be going for triples either.”

10. LP capital will overshoot, then flee at exactly the wrong moment

• Rich’s cycle read: there’s an optimal amount of capital relative to the opportunity set, and stay-private-longer genuinely expanded the set — but “in financial markets things tend to overshoot, especially when the indicators of success are lagging, and venture is the most lagging market.” LPs steer on trailing 10-year returns and will withdraw “probably just at the point when they should be investing” — he’d guess a significant change in capital availability within five years.
• His timing tell from 2009-10: “It would not be a great time to invest in venture until people spit at you when you mention the word.” When LPs said “the last good fund you have was a ‘96 vintage — get out of my office,” that was the moment to do nothing but venture. The inverse holds now.
• Jason’s heuristic: over a ~7-year window, exits should roughly equal new venture capital in — “and it’s been a hot minute for IPOs.” Rich: people don’t stop doing stupid [stuff] because they figure it out — “they stop when there’s no more money to do stupid [stuff].” Ben Stein: “If something in economics can’t go on forever, it will stop.” Rich’s corollary: “Until idiocy has to stop, it will go on.” The test is whether the Stripe/Databricks/OpenAI IPO cavalry arrives fast enough.
• Harry’s LP reality: two LPs a week managing $350-500M put $100M into Index, Excel, Founders Fund, Sequoia and ask where the other $250M goes. Rich: build a portfolio that won’t match the best fund but “comfortably outperforms the public markets” — the alternative is small-cap publics at 11%.

11. Staying private for longer is a public-policy failure — savers eat the fees

• Clari (likely — “Cler”) and StubHub pushed IPOs; the Collisons ask “why do I need some analyst at a bank to tell me about my margins?” Rory: they’re correct — and that’s the failure. It’s now more attractive to take capital from GPs paid 2-and-20 than from Fidelity Growth at 70bps for the same investment: “We have defaulted to the higher-priced capital alternative, which is absurd.”
• The saver’s math: Stripe compounds 15% gross either way. Public mutual fund: 14.3% net. Private venture vehicle: ~10% net after fees and carry. “The ordinary investors of America are either not getting the good assets or getting them at massively higher fees — a monstrously stupid outcome.”
• Why it persists: “being public is a pain in the ass” and “being private is cheap and easy money” — both must change. A top founder told Harry “there is no really significant reason for any great company to go public today.” Rory’s endgame: late-stage privates eventually underperform equivalent publics by exactly the fee load, capital reallocates, and the free money stops — “it’ll take a long [time].”

12. SSI at $32B: fund the secret recipe — but don’t count on the pref

• Rich’s logic on the “$32 billion, $2 billion raised, supposedly no product” round (likely Safe Superintelligence): fast-forward three years and every foundation-model company not populated by OpenAI people hasn’t done great, while Anthropic — which was — has, “with the exception of Grok (likely — ‘groth’), which is astonishing.” So: “You got the guy who invented the secret recipe. Why not?” What the model is worth once cracked is “a totally separate discussion.”
• Rich goes categorical: “There is zero chance this does not get bought for at least the pref. It’s [expletive] Ilia. Microsoft will buy him for 10 billion tomorrow, provided the government lets them.”
• Harry’s dagger — can you actually count on the liquidation preference? Rich: “Brutal comment, and you’re quite correct” — Delaware arcana offers ways to not honor it. Rory on today’s M&A: “Every acquirer is looking for ways to get around the VC preference stack… we don’t give a rat’s ass in corp dev what the certificate of incorporation says — side deals, back deals, nothing to the VCs.” Rory admits doing the same when his late-stage companies buy early-stage ones — though so far “the observed fact is investors have made money in sideways sales.”

13. Hot-round terms: every box checked, and risk quietly shifts to founders

• Rich’s field report from his hottest companies: later-stage investors now put everything in the term sheet to win — “maximum secondary, maximum refresh, even cram down the prior investors… all the boxes checked straight out of the gate.” Deals close in one day, and “if you check so many boxes, there’s an argument the valuation doesn’t even matter.”
• The growth-stage play: “Sell five [percent], we’ll give you seven” — a pre-approved equity refresh exceeding the founder’s secondary. Rory finds it “nauseating — you’re effectively replacing the comp committee of the company you’re investing in” — but has lost a deal by refusing: “bad money drives out good, and bad habits drive out good habits.”
• Harry’s structural worry: growth funds treat outcome sizes as equiprobable — pay $3B for a $2B company on the theory it’s a $10B company. “If I stuff Rory with $200 million before Rory’s ready, that $10B outcome will be a $4B outcome.” (Rory: “I’m always ready for you to stuff me with $200 million — let there be no ambiguity.”)

14. Deel/Rippling and the bezel: when the tide goes out, GAAP is what’s left

• On the alleged Deel spy inside Rippling ($5,000 a month; “clearly this guy went into the toilet and flushed his [phone]”): Rory — flagging his bias as an investor in Papaya, same space — says this crosses from driven into “industrial espionage… no CEO and no company can survive criminal liability”; if it turns criminal, a payroll customer “probably has to find a new payroll provider.”
• Mostly, though, it won’t take the company down — “2% at the margin,” a weapon for rival sales teams, because “you know how much work it is to change payroll providers?” The public-company playbook: termination drafted “before the attorneys stopped speaking,” interim CEO, ex-SEC lawyer, “lose a year.” Meanwhile Rippling raising at $18B: “if only for his cunning and acumen, you’d want to give him money.”
• Jason isn’t shocked: his survey of 2,000 B2B sellers found “93% said they lied to win deals” — “there’s going to be a hundred of these… revealed when the tide goes out.” Rory’s frame is Galbraith’s “bezel” from The Great Crash: at any moment there’s an amount of embezzlement outstanding; booms grow it because nobody looks, and it surfaces when the tide goes out.
• Hence the shift in diligence: “ARR is a made-up number and GAAP is a fact.” ARR has more signal about the future but more variance about correctness — and today’s experimental ARR is “neither A nor R — it may or may not be revenue, definitely doesn’t recur, and in no way is it annual if everyone can get out after a month.”

15. Buy or sell: OpenAI at $300B splits the table; Cursor at $10B is a durability bet

• OpenAI at $300B: both guests pass — “north of 100, so I’m out… all my decisions are bad” — while Harry is “the opposite of you two: I would buy the [expletive] out of this. Escape velocity reached.”
• Cursor at $10B: Rory notes the multiples are “pretty low, relatively speaking” — he paid ~10x revenue for Lovable — and “there are singularly amazing companies in every generation… you’re just so glad you bought it at any price.” Rich’s fork: if it’s durable SaaS — “coming up on a billion” with “140 or 200% NRR on paper,” a B2B company with a massive moat — “it’s a pretty good deal.”
• The catch, “the question of the ages”: is the revenue durable? Everyone Rich talks to switched IDEs in a week (“oh, Windsurf is cool” — his portfolio companies and his son switch back and forth). “If it’s SaaS, take my money — I just wish I had 500 million.” Closing bull math on OpenAI: replace all of Google and the market cap gives “plus or minus a little over a trillion — 3-4x from here in three or four years. Is that the best 3-4x you can do? I don’t know.”

Verification Notes

• The second OpenAI buy-or-not response is speaker-ambiguous in the raw captions.