Why Margins Don't Matter for Early-Stage Startups | Gili Raanan

Summary

• Venture doesn’t work — by design. Gili’s opener: “The venture business as a whole doesn’t work. It shouldn’t work” — returns concentrate in a handful of franchises (Sequoia, Andreessen, Benchmark, Greylock, Lightspeed), and given today’s capital inflows he expects “some serious catastrophe for many of the players.” An LP who spread their venture allocation evenly “wouldn’t sleep well at night.”
• The seed math is deteriorating: ~350–400 new cybersecurity teams get funded every year, the hit rate is “one out of 150, maybe two out of 150,” and Israel — roughly 40% of the global market — minted just two unicorns in 2025 and one in 2024, versus the 2021 outlier of around seven that “changed the mindset of investors.” With seed entries at “150x and 100x ARR” versus a $15M post for Assaf Rappaport’s Adallom in 2012 that Gili qualified with “if I’m not wrong,” “a lot of that cash that’s flowing into the market would be wasted.”
• Harry’s boomer challenge — CrowdStrike/Palo Alto-sized outcomes justify bigger entry prices — is accepted “with all humbleness” but doesn’t move him: “It would not change the probability facts around this game… we analyze smoke,” because founders reinvent product and market within weeks of a seed check.
• Fast growth is DNA: likely Wiz went $1M→$2M→$8M→$24M by quarter in its first year of selling software, and the bar for greatness — 4x/4x/3x/3x on new ARR, 144x over five years — hasn’t moved in the AI era. He’s confident someone will prove “Wiz was a slug” within five years.
• Gross margins matter, but wait until 2029: he has never once discussed gross margins with an early-stage portfolio company — “gross margins are important. Let’s talk about it in 2029” — though he suspects gross margins will continue to matter for AI too. Overcapitalized founders never worry him: “I’m not in a business of babysitting founders.”
• IPO is a branding event, not liquidity — “it’s hell for liquidity” — so the private-market extension is “functional and sustainable,” and recurring employee-tender secondaries (Cyberstarts’ employee liquidity fund, first program just done with Ayera, as heard) are the structural fix for fully-vested talent walking out. His confessed mistake: “I regret I sold every single share at Wiz.”
• To investors whose frameworks (rule of 40, triple-triple-double-double) are kind of out the window, the closing counsel: “Learn as much as you can from old farts like myself, but at the end of the day use your guts. Nobody knows better than you do.” Early-stage investing is “the science of greed” — “we need to be selfish, and we need to be greedy. Those are good traits.”

Deep dive

1. Venture doesn’t work — by design, and the math is getting worse

• Gili’s opening claim: “The venture business as a whole doesn’t work. It shouldn’t work” — returns aren’t divided equally between players, “otherwise it would be too easy… none of us would be playing.” The set of durable winners is “super small,” and given the money now flooding in, “I think it’s going to end up with some serious catastrophe for many of the players.” An LP with an evenly distributed venture allocation “wouldn’t sleep well at night.”
• His home-market math: ~350–400 new cyber teams funded per year across the US, Israel, and a little Europe — about 4,000 startups over the past decade, probably 4–5,000 in the next. Yet Israel, roughly 40% of the global market, minted two unicorns in 2025, one in 2024, and “two or one” every year back to 2022; only 2021’s outlier of around seven “changed the mindset of investors.” Hit rate: “one out of 150, maybe two.”
• With entry prices climbing — his first check to Assaf Rappaport’s Adallom in 2012 was at a $15M post, qualified with “if I’m not wrong” — “the market is not balanced… a lot of that cash that’s flowing into the market would be wasted.” His founder-facing corollary: “Pick your financing partners more wisely. The probabilities are not working in your favor. They’re working against you.”
• Harry’s pushback — “do you think you’re being a boomer?” — labor displacement and CrowdStrike/Palo Alto-scale outcomes justify paying more at entry. Gili accepts it “with all humbleness,” then refuses to move: “It would not change the probability facts around this game. Venture is a game… mostly we analyze smoke” — founders change product and market within weeks of the check.

2. Some mega funds can do well — greed is the job

• On $6–10B pools at Andreessen and peers: funds with “the tradition, the textbook, the guardrails” would continue to do well, and he’d personally invest in them. The opportunity is real, fast growers need more cash than before, and “cloud, code, and AI… would not change that materially” in the next few years. His concern isn’t fund size but entry prices eventually “limiting innovation… because disappointment would show up.”
• The self-description worth keeping: “We are exercising the science of greed. We need to be selfish, and we need to be greedy. Those are good traits for an early-stage investor.” When a price is inflated and the deal is “essentially a bet on a team,” he gets more skeptical — though whether he passes “depends on many other factors.”

3. Fast growth is DNA — it doesn’t fade

• Growth rates are “the most important indicators for a healthy business,” and the job is sensing engineered versus organic. Once a company grows super fast year over year, “it becomes part of their DNA… there needs to be a significant external event to slow them down.”
• The specimen: likely Wiz’s first year of selling software went $1M → $2M → $8M → $24M by quarter (2020 into early 2021). Having seen Palo Alto Networks’ and ServiceNow’s numbers inside Sequoia, his verdict: “this is an insane pace that the likely Wiz demonstrated.”
• Harry’s zig-zag belief gets a counter: a portfolio company heard as “Sierra” and likely Cyera sold ~$500K, then $1M, then literally zero for two quarters — “as an investor, you look at yourself and say, okay, I really fucked up” — then the founders retooled and sold $12M of new business in the next 12 months. Whatever makes a company move fast “would not simply fade away.”

4. The science of exceptions: No Name vs. Island

• Harry’s lesson — most companies plateau because markets are shallower and more crowded than believed — gets “I don’t think you are wrong,” then two contrary specimens from the same 2019 vintage.
• No Name did ~$3M then $15M — and slowed, because API security was “a niche segment within application security”; reinventing the market vision proved too hard and the business sold to Akamai for ~half a billion dollars. Depth caps even great execution.
• Island sells an enterprise browser — in 2019 the number of CISOs asking for one “equals the number of users that told the market they need an iPhone.” Today it’s a $5B company growing very fast in a market it defined, with banks and Fortune 100s picking it over Google and Microsoft: “you’re essentially competing with free.” The takeaway: “We are exercising the science of exceptions… if you just take those lessons and apply them linearly, I think it would be very hard for you.”

5. Cash doesn’t worry Gili; margins can wait until 2029

• On capital piling into winners and defocusing them: “I’m never worried about that. Never.” Building takes enormous cash — “if we don’t need it this year, we need it next year” — and to Harry’s young-founder-defocus scenario: “I’m not in a business of babysitting founders.” If you trust a team with the nation’s most sensitive information, “you can’t handle the idea that they have some extra cushion in the bank?”
• The real distinction is yield, not cash: 10 cents of new ARR per sales-and-marketing dollar is “a horrible business”; with product-market fit and decent-plus execution, 65 cents growing into 80 cents on the dollar is fine even if it’s not 140 — “why would you care that you have another extra $200 in the bank?”
• On AI eroding gross margins via inference spend, an honest non-answer: “I don’t think we have seen enough healthy profitable AI businesses” to know the vital signs. In cybersecurity margins clearly matter — yet he never raises them with early-stage companies: “Gross margins are important. Let’s talk about it in 2029.” He suspects they’ll end up mattering for AI too.
• The greatness bar hasn’t moved despite Lovable/Legora/Harvey-era growth: 4x, 4x, 3x, 3x on new ARR over the first five years — 144x — turns $1M of year-one new ARR into $144M booked in year five. Beating it is welcome (“5 to 50 to 200 — please do”), and even 1-4-16-48 makes “a very, very nice company — maybe not the most iconic ever.” In five years, he expects a team to prove “Wiz was a slug.”

6. Multiples are growth expectations; IPO is branding; secondaries are the fix

• Harry’s public book has gone from green to red — Monday at ~1.5x, Wix at ~2.5x announcing a big buyback at a $4B cap. Gili’s hedged read: “the multiplier is just the market anticipation for your growth rate” — his guess is that markets may be pricing in “autonomous programs” eating these businesses, and if growth persists regardless, multiples rebound. Kept as hedged: “I’m not sure and I’m not confident about what I’m saying.”
• On the likely Stripe/Canva-style private extension: “functional and sustainable,” because “going public is not a financial event. It’s a branding event… it’s hell for liquidity” — shackles, restrictions, the opposite of a liquidity event, paid for the long-term value of telling customers and employees “I’m here to stay.”
• Secondaries as talent retention: after four or five years your best engineers are fully vested with most of their family’s wealth in one stock, and diversification logic “actually forces them out of the company.” Cyberstarts’ antidote is an employee liquidity fund underwriting a recurring annual tender — first program just done with Ayera (as heard), “many millions of dollars” across a few hundred employees, priced in an ongoing process with management.
• The confession: “I regret I sold every single share at Wiz.” As a new GP he wanted to show LPs real liquidity; they cheered — except one or two, one of whom called to say “I’m not investing here to diversify… I actually like to take more risks.” He still thinks it was right for Cyberstarts at the time.

7. The craft: strengths over gaps, guts over frameworks

• Roughly 50 zero-to-one journeys in, his verdict on the job: “in many ways it’s a terrible profession — show me another profession where you show up to work every day for five years and have no idea if you’re doing any good.” At Sequoia, under Doug Leone, Michael Moritz, Jim Goetz and Pat Grady: “I’m the shittiest investor in this room… and the next day I’m still the shittiest investor. It takes a lot of greed and determination to keep going.”
• Partnership lesson for Harry: don’t bridge new partners to your recipe — “on their weaknesses, at best they can be as good as the market”; let each play their exceptional strengths, where they create “real greatness.” His biggest change of mind in 12 months: founder chemistry — back the teams you have the most chemistry with; the founding relationship matters “extremely” (his test: were they roommates, did they survive challenges together).
• He has lost deals in the last five years and missed a seed announcement “maybe once or twice over the past 8 years,” but rejects the Pat Grady every-miss-counts doctrine Harry quoted: “We are always as good as our next investment… let’s focus on our own thing. You are not going to win every battle.”
• Closing counsel to the generation whose frameworks (rule of 40, triple-triple-double-double) are kind of out the window: “Learn as much as you can from old farts like myself, but at the end of the day use your guts. Nobody knows better than you do.”