Sequoia's Bek: inside the IC, the SpaceX one-vote, and reading founders

Summary

• Bek’s core myth-bust: “Everyone thinks that we’re just waiting for the phone to ring for the next Anthropic to call us to invest. That’s completely false. Everyone at Sequoia is a hunter.” The proof case is Citadel Securities — Ken Griffin had never taken outside capital, and Sequoia got in only because partner Constantin Guler had cultivated the relationship since he was a student and “never gave up and just kept asking.”
• The best investments across every Sequoia fund are the ones where the sponsor had the highest conviction — not the highest ownership or cleverest structuring. Sean Maguire’s SpaceX proposal drew a vote of one (“I didn’t even know it was on the scale”), yet he forced the whole partnership to fly out, a small check became a big one, and it’s now “one of the best investments in the history of the firm.” Sponsors can green-light over the partnership’s objections — “if you press green and it’s a bad investment, we’ll see how long you stick around.”
• On repricing your own passes: Sequoia underestimated Anthropic early, then paid up $2.5B later — “revisiting our priors” because “the human brain’s just not very good at dealing with exponentials.” Stebbings’ framing lands: “$1 billion could just be the new Series A” — same blunt multiple as the old $50M-post-to-$1B path, now $1B-to-$20B.
• Bek is bearish on new AI labs: backing one now “you’re basically investing in the Quora, in the StumbleUpon, when Facebook came about” — unless it’s an N-of-one founder on a genuinely different architecture, citing Sequoia’s large seed into Ineffable with David Silva (likely David Silver) in the UK. He also calls legal the most over-funded category (“the winner is already in existence” — Harvey) and BCI the most under-funded (“where all the smart kids are going”).
• His agents-as-the-new-customer thesis: agent traffic has already reached parity with human traffic, Cloudflare projects 1,000x human traffic in five years, and businesses need a “bits-perfect platform that’s good at converting agents” rather than pixel-perfect websites. Agents carry pre- and post-training biases (defaulting to Cloudflare, Vercel), hedge funds are already buying data on agent decision-making because that may influence affected stock prices, and AEO isn’t just a new category but “a parallel economy for agents.”
• The viral services prediction, clarified: “the next trillion-dollar company will be a software company that masquerades as a service business” — capturing the $6 of services spend for every $1 of tooling. Customer support is already there: an “autopilot category” with $1B in ARR, where Sierra resolves airline tickets at a fifth of the ~$50 human cost and charges on outcomes. But he won’t back services-first companies (medium-high conviction): “you’re just not gonna get frontier talent wanting to work for an old service business.”
• Founder-reading tradecraft: open up first to earn vulnerability, then “ask why five times and you’ll get to the bottom of it” — the method that exposed a fraudulent founder claiming $0 to $7M ARR in six months. Weight “distance traveled” (the Polytechnique grad from care homes vs. the PE tycoon’s son), calibrate references by country (German customers’ NPS of 7 gets one-to-two points added, American reads get docked), and heed Alfred Lin: “Do not mistake an outlier operator for an outlier founder.”
• The best story in the episode is a miss: Bek lost Revolut at seed as his first-ever deal, then got into an SPV personally — funded 50/50 by his mother. Entry around $180-200M against a latest valuation over $100B; his mom sold most of her shares and retired at 74, so “they could only hire the second-best investor in the Bek family.”

Deep dive

1. Sequoia is a team of hunters, and the Citadel deal proves it

• Bek’s day-one story sets the culture: arriving jet-lagged at 4:30-5:00 AM, he found Doug Leone already inside — “I’ve already taken my first call” — before walking off delighted. Bek’s read: Sequoia doesn’t pressure people into that; “we just hire people who are built like that.”
• The biggest outside misconception: “Everyone thinks that we’re just waiting for the phone to ring for the next Anthropic to call us to invest. That’s completely false. Everyone at Sequoia is a hunter.” The early team was 11 people — “basically what a football team is, and everyone’s just scoring on the field” — with the expectation to “behave exceptionally well as individuals, but win as a team.”
• The illustration: Citadel Securities had never taken outside capital. Constantin Guler built a relationship with Ken Griffin from his student days, was mentored by him for years, “and just kept asking, ‘Can we invest? Can we invest?’ Until Ken kindly said yes.”

2. Revisiting priors: paying multiples for what you once passed on

• Stebbings flags the hardest discipline in venture — turning down a company, then “having the mental flexibility to pay multiples of it later and get over your own ego” — as Sequoia underestimated Anthropic early and later made a $2.5B investment. Bek’s term is “revisiting our priors”: three years into AI, “we’ve seen that exponential starting to play out, and suddenly we realize the human brain’s just not very good at dealing with exponentials. We can think very well linearly, but not exponentially.”
• Stebbings’ reframing of outcome inflation: “$1 billion could just be the new Series A” — the old game was $50M post hoping for $1B; now you enter at $1B and it becomes $20B, “same blunt multiple.” Bek’s caveat: “picking has never been harder because you have just so much more volume of companies.”
• On new frontier AI labs, Bek is blunt and flags it’s not the house view: “if you’re gonna invest in a new AI lab, you’re basically investing in the Quora, in the StumbleUpon, when Facebook came about.” The only exception is an N-of-one founder pursuing a different architecture — Sequoia’s large seed into Ineffable with David Silva/Silver in the UK — because “they’re not trying to do the exact same thing but better. They’re trying to be different.”

3. Ownership concentration survives because the model is co-founding, not indexing

• Asked whether expanding outcomes make Sequoia less ownership-centric, Bek’s answer is a flat “No. No. No” — and the binding constraint isn’t capital, it’s time: “In your career, you can make 20 investments… I partner with two, three founders a year.” His pitch to founders: “I’m gonna be basically their co-founder. They decide how to run the business, but I sit in the passenger seat.”
• The proof of what that looks like: for Rillet, “we met with 17 public company CFOs since the start of the year. Some of them have become customers. How do you do that when you have 200 companies with 2% in each of them? It just doesn’t work.”
• On the firm’s talent map: best sourcer is Dean Meyer in Tel Aviv — ex-professional footballer with “the competitive juices of Messi coupled with the technical depth of someone who’s been working in tech his whole career.” Best picker is Luciana Lisandru, who brought Bek in: Deliveroo, Framer, Pennylane, Stark — “banger after banger. And if you look at the pattern, there’s no pattern.”

4. Conviction beats consensus: the SpaceX one-vote and the Airbnb seed

• Sequoia votes deals on a numeric scale, and when Sean Maguire brought SpaceX, “I think someone voted a one… I didn’t even know we could do one.” Maguire refused to give up, flew the partnership out to see it, and a small check compounded into “one of the best investments in the history of the firm.”
• The recurring lesson from every offsite’s fund-return review: “every time we try to be cute to look at the numbers… the best investments in all the funds are always the companies where the sponsor had the highest conviction.” Where small dollars become big dollars, deals “have to be controversial” — Airbnb’s seed, after most firms passed Chesky, was among the highest money-on-money returns Sequoia ever made, despite “sleeping on air mattresses on people’s floor” sounding “like a pretty bad idea.”
• Bek pushes back on the “Sequoia just pays up” narrative: the firm partners early and often finds “capital that’s happy to pay a premium to that valuation” behind them. On the viral tranche-round claim, he’s seen it “a handful of times” and calls it supply and demand: “Why would they not command premiums after someone has invested?”

5. Greenfield vs. replacement markets — and Harry’s secondaries counter

• Bek’s strong view on today’s vertical growth curves (“I don’t know if it’ll be three years, maybe five, but this will come back”): people conflate new markets with replacement markets. AI-native CRM still has to displace a system of record; greenfield agent companies face no incumbent — “right now it’s apples and oranges, and no one’s really paying attention to that.” When greenfield saturates and hits replacement dynamics, the comparisons will normalize.
• Stebbings’ opportunity-cost pushback — capital should chase the faster-growing greenfield — gets a direct “I disagree”: outcomes crystallize in ten-plus years, the best companies stay private longer, “but you’re making a decision that will impact the business over the next three years,” and the biggest outcomes will “almost guaranteed” sit in maturing markets.
• Harry’s counter is structural: a frothier, more liquid secondary market than ever lets a smaller fund “sell into a liquid secondary market at a very exuberant price… in a way that you can’t.” Bek’s rebuttal, with a grin: “You have a 500 million fund. That’s a lot of money to do secondaries that have multiples.”

6. Inside the IC: async memos, founders pitching, and front-stabbing

• Well into its fifth decade of Monday ICs, Sequoia is experimenting: memos circulate and every partner contributes asynchronously, with anyone able to call a live IC. The logic — “an IC is a great format for fast thinking. Speaking asynchronously is great for slow thinking, and so if you can get the benefit of both, you’re hopefully gonna make better decisions.”
• Founders still pitch the full IC; Bek describes about 12 people in the early team and roughly the same number in growth. He deliberately doesn’t over-prep founders: “You need to see the essence of the people. If you give them a script, no one’s gonna see what you’re seeing.” A bombed IC is signal, not verdict: “Did you have questions about the founder being commercial? If you bombed the IC, maybe your questions were well-founded.”
• Votes are submitted before discussion, re-cast after, and fully visible — necessary because “it can’t be Harry’s investment or Julien’s investment. It needs to be a Sequoia investment,” so every partner opens their network in a heartbeat. The sponsor can still press green against the room: “if you press green and it’s a bad investment, we’ll see how long you stick around.” Culture-wise, Bek embraces “front stabbing” — and defends heated dissent as “a feature, not a bug. You want people to come in with courage. If they don’t have courage, they won’t take risk and we’ll have mediocre investments.”
• The inverse danger: when everyone scores a seven or eight, “quite dangerous” — the best founders “retrofit the narrative that they think is going to land.” Sequoia assigns a devil’s advocate to “write the pre-mortem of that investment before we make it.”

7. Reading founders: vulnerability first, then ask why five times

• Bek’s method for a 30-minute read — in a job where the mistakes that kill you are “omission mistakes, not commission mistakes” — is to open up first: he tells founders about the split household, the week with his mom overlooking Lake Geneva alternating with a mattress in his dad’s one-bedroom apartment, his mother beating cancer when he was six. “Otherwise, you’re just in a transaction all the time.”
• The fraud story is the counter-case: a founder claiming “$0 to $7M of ARR in basically six months,” with a too-perfect origin story of turning down Stanford. Repeated whys produced accelerating tempo and nervous body language; days later, en route to the airport at 5 AM, the founder canceled on a “family emergency,” and by that evening investors in the company confirmed he’d “come out as a fraud.” Bek messaged competitors to warn them. The lesson: “ask why five times and you’ll get to the bottom of it.”
• On arrogance, he reaches for Don Valentine’s two-by-two of founders-you-like versus founders-who-make-money: “your job is to figure out in which part of the quadrant we make money.” Arrogance “might be the cost of their spike” — the red flag is when it hides the absence of one. His own misread: lunch with Anton Osika before Lovable — “I just didn’t see it… I was not intentional about asking the right questions.”
• Two calibration tools: country adjustment — German Mittelstand customers giving Tacto a consistent NPS of 7 (“Because we can always do better. Very German… if they’re French or German, you add one or two points; Americans, subtract one or two”) — and “distance traveled”: of two Polytechnique founders, the PE tycoon’s son and the one abandoned at birth who grew up in care homes reveal completely different trajectories. “Your job is just to figure out if they’re gonna continue on that path.”

8. One lesson each from Doug, Pat, Alfred and Shaun

• Doug Leone’s interview weapon: after “Who is your best reference, and why?” comes “Who would be your worst reference, and why?” — and founders answer honestly. Bek then actually calls the worst references: “I’m not looking for perfection. I’m just looking for clarity.” Related warning from Shaun Maguire’s ELO framework: a 2400-rated chess player can identify another outlier in ten moves; a 2000 player can’t tell the difference — “you wanna ask exceptional people if someone is exceptional, not good enough people.”
• Pat Grady’s vector framework — people are direction times magnitude — plus the humility Stebbings recalls: every company that goes public, Sequoia saw at some point, “that just shows you how many we’ve missed.” The sentence every partner writes on day one, now printed on the wall: “We are only as good as our next investment.”
• Alfred Lin’s latest: “Do not mistake an outlier operator for an outlier founder” — which Stebbings ties to today’s CV trap of gold-plated OpenAI/DeepMind résumés. Maguire’s second framework: beyond IQ and EQ sit judgment and “political quotient,” and “judgment is actually more important than IQ, and PQ is more important than EQ.”

9. Agents are the new customer — a parallel economy, not a feature

• The thesis: three years into AI, agent traffic has hit parity with human traffic, and Cloudflare (cited that morning) projects 1,000x human traffic within five years. Twenty years of optimizing “a pixel-perfect website that’s amazing at converting humans” now needs a counterpart: “a bits-perfect platform that’s good at converting agents.”
• The thinking-fast answer — UI to zero, no brand loyalty, race to the bottom — is wrong, per Bek. Agents “have biases in their pre-training… biases in their post-training,” already defaulting to Cloudflare and Vercel for hosting; “you have hedge funds who are buying data to understand how agents are making decisions because that may influence the stock price of these companies.” Portfolio company Profound is “the answer to SEO for the modern marketer” — but Bek insists AEO is “not just a new category, it’s a new economy.”
• On margins: no destruction where switching costs persist — databases carry “data gravity, enterprise controls… trust, you build it over time.” And the human/agent boundary shifts with capability: today humans stay in the loop for holidays, but “80% of the databases are written by agents, so why would humans have a say as the AIs become so good that they can pick better?” On infra-vs-apps, he rejects Harry’s infra preference: “You can hold opposing ideas in tension and still be correct because those truths will materialize at different times” — Fireworks is ripping now; Rillet compounds into stickiness later. Rillet’s “Project Iowa” — car washes and auction companies as its fastest-growing segment — matters because “you wanna participate in the real economy, not just in the AI economy.”

10. Services thesis: sell the outcome, capture the $6, keep software margins

• The clarified prediction Harry called “word wank”: “the next trillion-dollar company will be a software company that masquerades as a service business” — masquerading is essential; it cannot be a service company. The math: businesses spend $1 on tools for every $6 on services (QuickBooks at $2K vs. $15K to the accountant), and the question is which categories can capture the $6.
• Customer support is already there — an “autopilot category” with $1B in ARR selling outcomes. Sierra approaches an airline paying ~$50 per human-resolved ticket and offers resolution “for a fifth of the price,” starting as copilot and quickly going autopilot, collecting a fee on the outcome, not the tool. Harry’s constraint stands: this works where resolution is cleanly verifiable; sales and marketing attribution ambiguity keeps most categories out for now.
• The bridge concept: copilots sit inside the human judgment loop, and if they build the right product, “the judgment of today is the intelligence of tomorrow.” Humans don’t disappear — “you start with lots of humans, little AI, and you end up with lots of AI, little humans” — and the data shows record hiring of system integrators, forward-deployed engineers, and software engineers (Jevons paradox: “we can’t underestimate human creativity”).
• His medium-to-high-conviction no: he will not invest in a services business planning to become software, because “the best companies are able to concentrate talent, and you’re just not gonna get frontier talent wanting to work for an old service business that’s kind of transitioned into an AI company. They might have data, but not necessarily the data.” Traditional PE, meanwhile, may do fine buying those data-rich businesses — but “we wanna back the next trillion-dollar company. I think it’s unlikely that’s a private equity-backed company.”

11. Quickfire, the Revolut miss, and why Sequoia has no house view

• Rapid calls: legal is the most over-funded category (“the winner is already in existence” — Harvey, with the widest distribution; “I just don’t understand investors’ willingness to invest in the Nth competitor”), though Harry counters with the unbundling thesis via Solve Intelligence in IP law. Most under-funded: BCI — “that’s where all the smart kids are going.” Best agent company outside Sierra was probably Cursor — “the first company to really understand that you could post-train models and go deeper into the stack”; it later joined Sequoia’s portfolio. Non-negotiable founder trait: “Intensity… It’s too hard to build a big business.”
• The Revolut saga: two weeks into venture at ~20, Bek saw Nikolay and Daniel Dines pitch in an East London basement — “think about the concentration of EV in that day.” Nikolay was “the most obvious founder call I’ve had in my career”; Bek camped outside Canary Wharf, lost the round (to Index and Balderton, he recalls), then asked to invest personally via an SPV. Earning 30K and penniless, he called his mom, who offered the money 50/50. Entry at ~$180-200M; latest valuation over $100B. He’s never sold a share; his mom sold most of hers and retired at 74 — “they could only hire the second-best investor in the Bek family… She missed all the bad ones.” His haunting myth is related: telling Trade Republic’s Christian “Revolut is going to smoke you” — “I failed to understand it was not a winner-takes-all market.”
• On internal dissent as brand: “people love to say, ‘Sequoia believes X and Y,’ but we actually have very different opinions. There’s no house view on AI” — his services piece coexists with David Cahn’s $600 billion question and Pat and Sonja’s AGI framing. “We’re trying to invest in spiky people, so we have to be spiky ourselves.” His five-year excitement: today’s ~120 IQ AI reaching 500 IQ, finding “cures for your mom or my dad’s disease… things so transformative to humanity that it will make all these things we worry about today sound completely insignificant.”