Oren Zeev: 50% of Funds Will Go Out of Business & Why GPs Shouldn't Tell LPs Their Strategy

Summary

• At least 50% of venture funds either cannot raise or aren’t sure they can, Zeev says — less money is going to venture and a larger share of it to platforms. His barbell: be an Andreessen/Sequoia/Lightspeed-scale platform or a differentiated solo/boutique; the traditional five-six partner firm with nothing special is “worse off on both hands” — corporate to founders, but “you’re not Sequoia.”
• Today’s growth expectations are BS: “the math doesn’t change — 2 to the power of 5 was 32 before AI and after AI.” He’d back a 2x-growing company with healthy economics over a 3x with unhealthy ones all day long, and warns growth-only optimization breeds circular deals — “no value was created… but a perceived value was created” — a gray area “before I get to fraud” that will implode for some.
• Radical alignment as fund design: Zeev takes 30% carry but pays himself zero from management fees (he reinvests 100% into the fund), is the biggest LP in every one of his 11 funds at ~13-14%, and sees “not a shekel” before LPs get 100% of their money back — roughly 40% of the economics. The contrast: a $10B fund at 2% produces ~$2B in fees over 10 years, starting today, while carry arrives in 7-8 years — so many GPs optimize for raising the next fund, not returns.
• The incumbents-die narrative is thought-leader bait: Navan is “100% convinced” to be a huge AI beneficiary with “zero chance” of disruption, and is one of Zeev’s most concentrated positions — support-heavy gross margins that were ~50% three years ago are “dramatically better already,” with almost all support eventually done by AI. The moat is operational complexity, distribution, integration, regulation, and data — “technology is 5% of it… who has the most data? The incumbents.” With SaaS multiples lower than they’ve been in the past 10-12 years because the market can’t yet discern winners from victims, mispricing cuts both ways.
• Paper marks are a motivation test, not a methodology question: Sequoia has zero incentive to inflate; an insecure mid-tier fund “will find any excuse to keep prices up,” and accountants “always challenge the wrong things.” Today’s DPI obsession is a cycle that will turn — possibly via a “tsunami of liquidity” in 2026-27 from unprecedented-size IPOs in the works: SpaceX, Stripe, Databricks.
• He doesn’t sell secondaries — “everything that I can sell, I don’t want to sell” — but concedes the math can work: when Harry defended taking 3x now over 4.5x in 2-3 years, Zeev’s second-grade-math verdict was that 1.5x over three years with execution and IPO risk means “you should have sold.” Managers may sell to manufacture DPI for fundraising.
• On AI labor displacement he sides with “this feels a bit different” over it-always-takes-longer wisdom — “I’m excited because I’m going to make a lot of money, but I’m also nervous.” AI is “the biggest change ever in the history of humanity,” and it’s “the best time in history to be an investor” — while political unrest from the disenfranchised is “very, very risky to humanity.”

Deep dive

1. AI added one mandatory question to his checklist

• Asked whether the AI wave changed what he looks for, Zeev’s answer is “not so much, surprisingly” — the fundamentals are the same fundamentals. The one addition: every investment now must pass “is this company a likely beneficiary of AI?” A victim is an easy no — but “even if the answer is neutral, then still the answer is probably no.” Four years ago he never asked it; now it’s non-negotiable.
• His baseline remains contrarian avoidance: the best outcomes look weird or wrong at entry precisely because there won’t be 15, 20, or 100 startups doing the same thing — “some level of contrarian plus being right — that’s the ingredients typically of great outcomes.” His filter: “if everyone is doing something, it’s a reason not to do it, not a reason to do it.”
• Harry’s test case — AI customer support, the consensus “duh” market: Zeev confirms he stays out. One or two of the thousands will win, but “I don’t trust my intuition enough to know which one of the thousand is going to be successful.”

2. The incumbents-die narrative is provocation, not analysis — Navan as exhibit A

• Zeev is “100% convinced there’s zero chance we get disrupted by AI and 100% chance we’re huge beneficiaries” at Navan, one of his best and most concentrated positions. He is not sure the market sees it that way for now: SaaS multiples are lower than they’ve been in the past 10-12 years because investors rightly fear AI disruption but “are not yet at the point where they discern” the negatively from the positively impacted. Over time some incumbents will look expensive even with the discount — and others will prove to be beneficiaries.
• The evidence he can share: three years ago Navan’s gross margins were ~50%, all support cost; with AI they’re “dramatically better already” and improving, with almost all support eventually done by AI — and the “even more exciting” part is what AI does for customer experience.
• His disruption framework: simple software that someone can rewrite quickly and undercut is at risk. But the more operationally complex, distribution-heavy, integration-dependent, and regulated a business is, the harder it is to displace — “technology is 5% of it… who has the most data? The incumbents.” The all-incumbents-die notion is promoted by people “whose main motivation is to make provocative statements… and get attention as thought leaders.”

3. Growth expectations today are BS — compounding didn’t change

• Harry’s frustration: 1→5M ARR no longer excites the big funds at the B or C. Zeev doesn’t buy it: “the math doesn’t change. If you have a company that can double every year for the next 5 years, it’s going to be 32x what it is today — 2 to the power of 5 was 32 before AI and after AI.” The real questions are whether growth is sustainable and healthy.
• Live disagreement, worth keeping: a company he backs is at $20M ARR doubling to $40M with very healthy economics, and a respected investor flagged 100% growth as a problem. Harry sides with the investor; Zeev: “I think he’s dead wrong on this one” — the company leads its market with no faster competitor, and “there aren’t many $20 million companies that are doubling with very healthy economics… it’s just not so many of them.”
• The danger of growth-only thinking — “I’ve seen this movie many, many times”: it drives unsustainable behavior like circular deals — “I’ll buy your product for a million dollars and you buy my product for a million dollars… no value was created in this theoretical transaction, but a perceived value was created.” That’s still gray area, “before I get to fraud” — and for some it will implode.
• The carve-out: in Uber-vs-Lyft dynamics you don’t have the luxury of growing healthy — “you just have to play the game and hope for the best.” He simply looks for businesses where that isn’t the dynamic, and concedes to Harry that focusing on margins too early can cap upside — the answer depends on the competitive environment, “there’s no one solution for all.”

4. Judge the decision, not the outcome — “we are self-validation machines”

• His biggest mistakes came from thinking he was smarter than the market — turning down a payroll seed (“paychecks, ADP, come on”) from “this kid Alex” — but so did his biggest wins. He actively suppresses social proof and invests on his own conviction: being right 50% of the time is a great result “because if we lose something we only lose 1x our money; if we win, it could be 100x.”
• The follow-on loss he owns: a proptech company that went from $2M to $30M run rate, projecting $100M. He doubled down at a discount just before the late-2021/22 rate spike. He had stress-tested rising rates — but “what I modeled as the worst-case scenario was actually not as bad as the real scenario that happened.” The company didn’t survive.
• The lesson he draws is “actually, not much” — it’s a mistake to judge a decision by the outcome. Quoting Annie Duke’s poker framing: right decisions can lose the pot, and over time right decisions win. Her line he loves: “we humans are not truth seekers, we are self-validation machines” — like his old partner who never met a follow-on he didn’t like, because every new fact proved he’d been right all along.

5. 20% concentration, 12-month funds, and the 2021 confession

• His concentration limit is 20% of a fund in one company versus an industry standard of ~10% — and he argues GP-level diversification “gives nothing to LPs” since they hold multiple GPs anyway. “I’d rather be concentrated in the best deals I can find, because then when you have a winner, it really makes a difference.”
• On LPs’ complaint that he deploys too fast and inconsistently: “I’m going to do my thing, and if it works for them, fine. And if not, they can opt themselves out” — some have, “good people,” and it’s fine. He admits it’s the thing LPs like least about him, but “I don’t want to not invest in a company when I think it’s compelling just because I just made another one.”
• The honest look-back: in 2021 “every single deal I probably paid three or 4x what I should because that was the market” — so a fund that would have been a 5x becomes a 1.5x. One of his 11 funds will be “okay, not great.” He doesn’t believe in his or anyone’s ability to time the market; LPs get vintage diversification across his rapid succession of funds, not within one.
• He also over-sized: two bubble-era funds over $500M from 2021-22; his 2024 fund cut that size roughly in half; Fund 10 is about $250M, and he wants Fund 11 below $250M. An LP who pushed him hard two years ago to shrink got told “you can help by just not being in the next fund” — they stayed, and “in hindsight, I think they were right.”

6. Be a platform or be special — the messy middle is in trouble

• He 90% agrees with Harry’s barbell thesis. Either you’re a platform — Andreessen, Sequoia, Lightspeed — that “can do things smaller VCs cannot, including myself,” or you differentiate the other way. “I’m not trying to be better than Andreessen at Andreessen’s game — they’re going to beat me every time.” His edge: faster than anyone else, personal connection, and founders who specifically want a solo GP.
• The traditional five-six person partnership with nothing unique is trapped: “it still feels as a corporate to the founder… and on the other hand, you’re not Sequoia, so you’re not going to get the very best deals.”
• The headline call: “at least 50% of the funds today, and maybe more, either cannot raise or at least are not sure that they can raise, or they’re trying to stall and not test the market — and I think many of them are not going to be able to raise.” Less money is going to venture, and a larger percentage of it to platforms.

7. Marks are a character test, DPI is a cycle, and 2026 could reshuffle everything

• On the 90%-of-unicorns-aren’t problem: whether you can believe a VC’s numbers “is less dependent on the methodology… it’s more a function of the character, but even more so the motivation.” Sequoia can raise anytime, so it has zero incentive to inflate; a middle-of-the-road fund unsure of its raise “is going to find any excuse to keep the prices up.” Accountants are no guardrail — “they always challenge the wrong things.”
• After a 4-5 year liquidity drought, LPs have swung to DPI — “people hardly talked about it 3 years ago and now some LPs: it’s just DPI, we don’t believe anything.” He thinks that’s an understandable but blunt response to unreadable TVPIs — and “I do think it’s a cycle. It will change again.”
• The catalyst: “there could be a tsunami of liquidity in 2026-2027” from a host of unprecedented-size IPOs in the works — SpaceX, Stripe, Databricks and others — “and that would reshuffle the cards again.”
• On secondaries, he abstains by construction: “anything that I want to sell, I won’t be able to. And everything that I can sell, I don’t want to sell” — buyers aren’t stupid, they only buy what can double or triple, and demand a discount for it. Harry’s pushback — they sold something for 3x now over 4.5x in 2-3 years laden with execution and lock-up risk — earns Zeev’s concession via “second grade math”: that’s only 1.5x over three years with a lot of risk, so “if this is what you believe, you should have sold.” Managers who need it, he argues, may sell to show DPI and raise.

8. Zero fees, 30% carry — “100% substance, 0% appearance”

• His structure: a low management fee, 100% of it reinvested into the fund, and he pays himself nothing — “I don’t know any VC in the world that has zero income from the management fees.” He’s the biggest LP in every single fund (~13-14%, no other LP above 10%), takes 30% carry — “I’m 40-something percent of the economics” — and by construction gets paid nothing “before the LPs got 100% of the money back… they don’t see a shekel.” His summary: “I’m 100% substance, 0% appearance.”
• The big-fund misalignment he’s built against: a $10B fund at 2% produces $2B in fees over 10 years, starting today, while carry — even on doubling the fund — arrives in seven-eight years. Time-value-adjusted, fees beat carry, so “their whole thinking is: what do we need to do to raise the next fund” — including selling early to show DPI.
• The second misalignment is inside the partnership: younger partners “are first and foremost managing their career” — zero incentive to admit failure, every incentive to convince partners to roll the dice on a follow-on, “and even if not, they bought some time personally.” The larger the partnership, the worse it gets. In his case: “it’s just me… there’s zero conflict.”

9. “I only have one rule, and that rule is that I have no rules”

• Why GPs shouldn’t tell LPs their strategy: “if I tell LPs something, I would feel too committed to that specific strategy” when a situation demands flexibility. “So I tell LPs I only have one rule, and that rule is that I have no rules” — and it means never having to explain later why you didn’t do what you said.
• The proof case — the AI company he calls “the cart”: two exceptional founders, no idea yet, closing $3M from angels in 24 hours on an uncapped SAFE. He normally does no SAFEs; here he took $1.5M (the angels cut back 50%), insisted the SAFE be capped — “luckily,” given the next round’s much higher valuation — and ended at ~5%. He never got to increase it because the company turned profitable “very, very, very quickly,” taking Sequoia (Sean Maguire) and Benchmark money only because it wanted those names.
• On Harry’s claim that Series A is now the worst insertion point (200x ARR, little progress, steep markup): “scratch the word ’today’ — it’s always been the case,” even 30 years ago. Round names are just names; the only question at the second round is whether progress is “a real signal of product-market fit or just noise.” Sharpest twist: no PMF signal after a year or two may mean the company is worth less than at seed — “at the seed you had the option value.”
• On preemptive rounds shoved in by platforms a month after a raise: founders should “take the money, but continue to behave as if you didn’t” — companies can be overfunded into loss of focus, though he can’t fault a founder for taking 50 million at a high valuation.

10. Misses, the anti-portfolio fallacy, and which side of the AI debate he’s on

• Against Jason Lamin’s claim that founders no longer want investor advice, Zeev isn’t feeling it — because he never forces advice and founders know he’ll back them even when he disagrees: “for them I’m a safe environment… like going to a psychotherapist.” Needing to convince someone kills receptive mode; knowing support is unconditional “disarms them.” Delivery matters too — come from know-it-all and founders (and GPs, per his own LP story) tune out even correct advice.
• His misses cut the other way: he has few, “which means that I’m not seeing a lot of the great ones — it’s not a good thing.” Deel he saw and “really didn’t see it.” From his Apax days: early Facebook was dead on arrival at the partnership, and Audible — his first big home run — he tried to take private after the stock dipped, couldn’t get approval, and it went to Amazon instead.
• Wiz is his lesson in humility about misses: his intro reached Assaf’s dormant Microsoft email, but honestly “I don’t think I would have got the deal anyway” — it belonged to the founder’s prior cyber investor, Gilly. Hence the anti-portfolio fallacy: “it doesn’t make sense that the same deal appears in 20 different anti-portfolios, because it’s not as if the 20 could have done it.”
• The closing exchange: on labor displacement he’s “excited because I’m going to make a lot of money, but I’m also nervous” — and when Harry poses experience’s “it always takes longer” against “this feels a bit different,” Zeev picks Harry’s side: “Yours — but I very much hope to be proven wrong on the concern side.” Still, “AI is the biggest change ever in the history of humanity… it’s the best time in history to be an investor” — with the caveat that political unrest from the disenfranchised is “very, very, very risky to humanity.”