Founders Fund former core partner Brian Singerman: Tear up the evaluation scorecard, seek extreme "spike" founders and heavy investment logic.

This interview is very worth digging into, and when viewed alongside Doug Leone's interview, one can see two completely different yet extremely top-tier VC worldviews. Doug Leone's core is: How do I identify extreme founders and enable a few super winners to compound over the long term? Brian Singerman goes a step further and asks: Since VC itself is also a competitive game, why should I look for "standard good investors" or "standard good founders" according to industry uniform rules? I should find my own asymmetric advantage and tilt the entire game in that direction. This is the most important aspect to understand about Singerman. What he discusses on the surface includes Spikes, Airbnb, Anduril, Founders Fund, GPx, but at its core is a complete set of: Asymmetric competition philosophy. Moreover, this interview with Uncapped's Jack Altman was just released on September 1, 2026. Singerman has already transitioned from Founders Fund General Partner to Partner Emeritus in December 2024 and co-founded GPx with Lee Linden of Quiet Capital in 2025. ──────────────── 1. First, let's change the title: the real keyword here is not "spikes," but "asymmetry." The titles you provided are already good. But if I were to distill it, I would recommend: From Airbnb's 30th investor to heavily investing in Anduril: Brian Singerman's "Asymmetric VC" philosophy. Or more thought-provoking: Don't be a perfect investor: How Brian Singerman wins in VC with "spikes, concentrated bets, and asymmetric advantages." Or even more aggressive: VC is not a draft, but a war: Brian Singerman's spike talent, concentrated betting, and the new experiment of GPx. Because the entire interview actually revolves around one question: What do you have that others do not? If not, even if every ability is: 8/10, you may still have no competitive advantage. ──────────────── 2. First, let's correct an identity issue: Brian Singerman is no longer a typical Founders Fund GP. He worked at Founders Fund for about 17 years, previously spent about four years as an engineer and executive at Google, where he began his angel investing. In December 2024, he transitioned to: Partner Emeritus. In 2025, he started the new GPx system. Therefore: "Former core partner of Founders Fund" is accurate. However, GPx was not officially completed in fundraising by the end of 2024. Public reports in July 2025 stated: It is raising over $500 million. As of now, public information clearly indicates that its investment model has already started, but I would not write down "over $500 million final closing size" unless the fund officially announces closing. ──────────────── 3. Singerman's biggest counterintuitive view on talent: Well-roundedness is not necessarily a virtue. Traditional hiring systems prefer: Leadership 8 points. Execution 8 points. IQ 8 points. Communication 8 points. Strategy 8 points. Emotional intelligence 8 points. In the end: The total score is very high. What is the biggest problem with this system? It naturally rewards "people with no obvious shortcomings." But entrepreneurship is not a civil service exam. VC is also not a civil service exam. The people who truly change the world often have: Some things 4/10, and another thing: 15/10. Singerman clearly stated in the interview that Founders Fund tried early on to summarize the so-called personality traits of a "standard FF founder," but later found that this checklist simply did not hold, so he completely abandoned the rubric and only looked for where a person "spikes" in certain dimensions. ──────────────── 4. Why might an average score of 8 lose to someone with "15 points + several 4 points"? Because success in entrepreneurship is not: An Additive Game. It is often: A Multiplicative Game. For example, a founder: Sales ability: 15/10. Technical skills: 5/10. Operations: 4/10. If he has: Strong self-awareness, he can hire: CTO, COO. In the end, it forms: 15 × 15 × 15. But a person: With all abilities at 8 points, and thinks: "I can do everything." Ends up authorizing nothing. Ultimately, the organizational ceiling may only be: 8. This is why: Self-awareness is sometimes more important than well-roundedness. ──────────────── 5. So spikes alone are not enough; what Singerman is really looking for is: Spike × Self-awareness × Leverage. All three must exist simultaneously. A person having super abilities: Is not enough. He must know: Where am I strong? Then know: Where am I weak? Finally know: How to redesign the game so that the outcome depends on my strengths? Singerman explicitly connects this to his gamer mindset: do not clash head-on in the strongest dimensions of others, but rather "tilt the game" to positions where you have an advantage. ──────────────── 6. This is why StarCraft is a very good business metaphor. The brilliance of "StarCraft" lies in: Terran, Protoss, Zerg are not: Three armies with completely identical attributes. They have: Asymmetric Balance. The advantages of the three races are different. Truly excellent players do not say: "I want to play Zerg as Terran." But rather: "Where does my race have asymmetric advantages?" The same goes for businesses. ──────────────── 7. This is actually a more practical version of Peter Thiel's "Competition is for Losers" from "Zero to One." Thiel's thought is: Do not enter a market where everyone competes using the same rules. Singerman's personal version is more like: Don't play a symmetric game. Suppose you are starting a business in AI coding. Competitors: OpenAI, Anthropic, Cursor, GitHub. If your strategy is simply: "My coding agent is also good." It's a dead end. You must ask: What spikes do I have that others cannot easily replicate? It could be: Specific industry data. Distribution. Developer Community. Enterprise Integration. Latency. Security. A specific vertical workflow. Otherwise: There is no reason to win. ──────────────── 8. This is actually a very important concept in entrepreneurial strategy: Choose the axis of competition. The biggest mistake of the weak: Accepting the competition dimensions defined by the strong. For example: Amazon compares logistics speed. You compare logistics speed with Amazon. Google compares search quality. You compare with Google on general search. OpenAI compares foundation models. A thirty-person startup also trains general models. These are all: Playing Their Game. Real entrepreneurs need to redefine: What exactly is the competition about? ──────────────── 9. This is also Singerman's current method for screening GPs. He is not asking: "Are you a smart investor?" Because: Almost all GPs who receive first-tier LP money are very smart. What he is really asking is: If this person were to face me at my peak in their own game, could they possibly defeat me? This is a very high standard. The core of what he said in the latest interview is: "I am looking for those who might defeat me in their own game at my peak." ──────────────── 10. This is completely opposite to looking for "the next Brian Singerman." This is the smartest aspect of GPx. Many successful investors ultimately look for: "People like me." Disaster. Because: They have already appeared. The edge of the next great investor may be completely different. Singerman's philosophy is instead: You do not have to play my game; you must have your own game. This is called: Manager-Founder Fit. Just like Founder-Market Fit. ──────────────── 11. This is why "we focus on Deep Tech" is not differentiation at all. Today, 100 funds all say: AI. Deep Tech. Defense. Climate. Founders first. Operator-led. Concentrated. Early conviction. These words: Are all commoditized. True differentiation is not: Strategy Description. But rather: Uncopyable Advantage. For example, a GP: Has worked in a specific vertical industry for 15 years. Knows every core founder. Or: Is one of the world's best Growth Hackers. Simultaneously helps multiple non-competing portfolio companies grow. Even if others know his strategy: They cannot replicate it. This is what is called edge. ──────────────── 12. This point is very similar to Buffett's "circle of competence" but more radical. Buffett: Stay within your circle of competence. Singerman: Not only stay within the circle of competence, but also: Move the arena into the capability circle. This is a level upgrade. ──────────────── Thirteen, Anduril is a perfect case for understanding "Spiky Team", but your materials are missing one person. Anduril does not have four co-founders. But five: Palmer Luckey Brian Schimpf Matt Grimm Trae Stephens and: Joseph Chen. Anduril's official sources and several early reports confirm that the five co-founded the company together. So the main text must include Joe Chen. ──────────────── Fourteen, but why does Singerman mainly talk about the other four? Because he is explaining: Complementary Spikes. The roles of these four people are extremely distinct. Palmer: Vision / Product imagination. Matt Grimm: Operations. Trae Stephens: Capital / Government / Defense ecosystem. Brian Schimpf: CEO / Engineering / Organizational judgment. And Joe Chen: adds military experience and hardware/technical capabilities. Early descriptions of the Anduril team by a16z also particularly emphasized this complementary structure. ──────────────── Fifteen, Palmer Luckey is not a traditional CEO-type founder. This is very interesting. Palmer's strength is: Product Imagination. He can come up with things that others might even find absurd. And then: actually make them. Oculus is the most typical case. Singerman was an early investor in Oculus; WIRED later directly described him as one of Oculus's first investors/early investors. But: Visionary and: CEO are not the same profession. ──────────────── Sixteen, this is the most common mistake in the tech industry. The person who invents the product: automatically becomes the CEO. Why? There is no logical necessity. A person may be the world's number one: Product Genius. But a CEO needs: organization, capital, conflict, recruitment, culture, priorities, government, customers, financing, board. This is a different profession. ──────────────── Seventeen, very few people can actually do both. For example: Elon Musk. Steve Jobs at certain stages. Jensen Huang. Max Levchin in certain scenarios. But this is: Extreme Outlier. Companies must not conclude from a few historical cases that: "Great founders must do everything themselves." This is a typical: Survivorship Bias. ──────────────── Eighteen, what Anduril is strongest at is not Palmer, but the "Founder Portfolio." This is where I would further elevate Singerman's viewpoint. One can understand a founding team as: Human Capital Portfolio. Just like a fund cannot put all its money in the same beta. The team also cannot have everyone possessing: the same abilities. If all five people are: great engineers, without: government relations, operations, capital, organizational capabilities, it won't work either. ──────────────── Nineteen, so what co-founders should really pursue is not "we get along particularly well" but rather: High Trust + Low Skill Overlap. Ideally: highly consistent values. highly different skills. This is the most beautiful combination. ──────────────── Twenty, Anduril just proves that this combination can create immense value. As of May 2026, Anduril completed $5 billion in financing, with a valuation reaching $61 billion, and revenue in 2025 already exceeding about $2.2 billion. Moreover, as of July, the company was also reported to be researching further financing, with a potential valuation approaching $100 billion; however, the company clearly stated that future financing had not been finalized, so $100 billion cannot be written as a completed valuation. This means: Founders Fund initially backed not just an ordinary startup. But one that could become: a new generation of large American defense contractors. ──────────────── Twenty-one, how much did Founders Fund lead Anduril's Seed round in 2017? About: $17.5M. Today: $61B confirmed valuation. This does not mean Founders Fund directly achieved 3500 times—there were a lot of subsequent investments, dilution, and different rounds, so it cannot be calculated this way. But it illustrates well: Power Law. ──────────────── Twenty-two, and Founders Fund made an even more astonishing move with Anduril by 2025. When Anduril raised $2.5 billion at a $30.5 billion valuation in 2025, Founders Fund invested: $1 billion. This was the largest single check in the fund's history. This is what Singerman referred to: Concentration. But it must be noted: This is completely different from the 2017 Seed investment. ──────────────── Twenty-three, this precisely illustrates that the most valuable ability of top VCs is not "getting it right the first time" but another ability: Update Conviction. In 2017: there was one judgment. In 2019: new data was obtained. In 2021: continued to obtain new data. In 2025: if the company is getting stronger, why should one be limited by "we have already invested"? This is: Bayesian Investing. ──────────────── Twenty-four, most investors have a serious psychological bias: Entry Price Ego. They like to say: "I was in at Seed." Because: it sounds impressive. So when they see a company that has already increased 20 times: they are unwilling to continue investing. Why? Because the new entry valuation: makes them feel "not worth it." But this is a wrong psychological anchor. The right question is: From today's price, how much future return is there? The past purchase price has nothing to do with this question. ──────────────── Twenty-five, Airbnb is a classic slap in the face of this ego. Founders Fund was not: the first, the second, the third, investor in Airbnb. According to Singerman himself in an interview: it was about: the 30th investor. Valuation: about: $2B. Investment: about: $150M. But according to his statement, Founders Fund ultimately became the second-largest institutional investor in terms of absolute amount earned from Airbnb. This belongs to Singerman's public statement, not the "investor return ranking" released by Airbnb after auditing, so it is best to note this in the main text. ──────────────── Twenty-six, why can the 30th investor still earn so much? Because: Multiple ≠ Dollars Returned. This is a concept that VCs easily confuse. Investor A: $1M. In the end: 100x. Earned: about $99M. Investor B: $150M. In the end: 20x. Value: $3B. Earned: about: $2.85B. Who has the prettier multiple? A. Who made more money? B. ──────────────── Twenty-seven, so VCs have two different scoreboards. MOIC how many times you earned. Absolute Profit how many dollars you earned. The two are not the same. Many early investors may have an astonishing: 100x. But due to low ownership, actual dollars returned may not be the largest. This is why: Position Size Matters. ──────────────── Twenty-eight, Airbnb's current scale further illustrates this point. As of September 3, 2026, Airbnb's market value is still around $110 billion. Looking back: $2B valuation was certainly very cheap. But at that time, it was absolutely not: "cheap Seed." What is truly difficult is: after missing the first 29 opportunities, can one still admit: others saw it right before me, but I should still bet heavily now. ──────────────── Twenty-nine, this requires a particularly scarce ability in the investment world: Low Ego Updating. Ordinary people's thinking: "I have already missed it." Top capital allocators: "Missing and whether to buy now are two independent questions." This is the same as stocks. You missed Nvidia: $10. $50. $100. Does not mean: $150 is definitely expensive. Judgment can only be based on: future cash flow / outcome. Cannot be based on: "Others bought it cheaper than me." ──────────────── Thirty, this is a very important investment rule. Someone else's cost basis is irrelevant to your forward return. Others' costs: do not determine your returns. Your returns depend on: today's price and future value. ──────────────── Thirty-one, so Singerman's statement "billions minus 5 million is still billions, minus 150 million is still billions" truly expresses Tail Outcome Math not saying: Price is not important. Of course, price is important. Rather, it is to say: In the face of a true: 100x Outcome Distribution, early price differences can sometimes be overshadowed by the magnitude of the final outcome. ──────────────── 32. But this thinking is very dangerous and should not be mislearned. It can easily turn into: "Good companies can be bought at any price." Wrong. If Airbnb at that time was not: $2B, but: $80B, the outcome would be completely different. So the correct formula should be: Expected Outcome / Entry Valuation × Ownership not: "If the company is good, buy it mindlessly." ──────────────── 33. Concentrated investment can also be easily mislearned. Singerman himself is very averse to: "Founders Fund is concentrated, so I will also concentrate." Because: Concentration is a result, not a strategic posture. You do not have: SpaceX, but forcibly put 20% of the fund into one company. This is not called brave. It is called: Uncompensated Risk. ──────────────── 34. True Concentration must meet two conditions. First: Extreme Asset Quality. You really encountered a: SpaceX, Palantir, Stripe, Anduril. Second: Extreme Calibration. You really have the ability to know: It is different from the ordinary "best portfolio company." The second point is the hardest. ──────────────── 35. Assume there are 20 companies in the fund. The best one: does not equal: should invest 20%. Because: Best in Portfolio ≠ Generational Company. You must have an external absolute standard. Otherwise: Every class has a top student, it does not mean: every class's top student is Einstein. ──────────────── 36. Singerman talks about the concentrated history of Founders Fund, which is actually very valuable. He recalls some typical concentrated bets of FF including: SpaceX, Palantir, Stripe, Airbnb, Stemcentrx, Anduril, and later OpenAI, etc. At the same time, he also clearly states: Not every fund must engage in extreme concentration. This is crucial. Because: "There are no worthy opportunities for concentration" is itself a legitimate answer. ──────────────── 37. This is one of the biggest disciplines in investing: Do Nothing is a Position. Many fund managers feel deployment pressure after receiving money. LP gave: $1B. Management fees start to be collected. So they feel: "Must invest." As a result, they lower their standards. Real excellent capital allocation: If there are no good opportunities: cash is also an asset. ──────────────── 38. This is exactly the same as Buffett. Buffett has historically often accumulated: huge cash. Not: "I don't know how to invest." But: The price and opportunities do not meet the standards. Although VC has stronger constraints on fund duration, the logic is the same: Capital Deployment Is Not a Moral Duty. ──────────────── 39. The internal mechanism of Founders Fund is also worth correcting: Peter Thiel is not the "CEO." The official position given to Peter Thiel by Founders Fund is: Partner. Not CEO. But Singerman clearly states in the interview: Founders Fund has never been an equal partnership like Benchmark, Peter has final authority/decision-making power on major issues. So the correct statement should be: "Founders Fund is not an egalitarian partnership; according to Singerman's recollection, Thiel has final decision-making power." Do not write: "Peter Thiel is the CEO of Founders Fund." ──────────────── 40. This is actually a very interesting organizational paradox. Founders Fund is very decentralized in: Investment Process. Different partners: have vastly different personalities. Investment methods vary. There is no unified founder rubric. But the final power structure: is not completely egalitarian. This is called: Decentralized Creativity + Centralized Sovereignty. Many great organizations have this structure. ──────────────── 41. Apple was also like this. Product teams: highly creative. But Jobs: Final Say. Amazon: many autonomous teams. But Bezos: has strong control over important principles. Founders Fund is somewhat similar: allows: extremely different investor spikes. But must have: a final organizational order. ──────────────── 42. Do not write that Founders Fund's "no Memo, no fixed Partner Meeting" will always be the case today. What Singerman said was: the early Founders Fund. They did indeed briefly try: investment memos, partner meetings, and then found it unsuitable for this early group of people and quickly abandoned it. This indicates that: organizational processes should serve talent, not talent serving processes. But this cannot lead to the conclusion that: "Today Founders Fund has no formal processes at all." There is no public evidence for this. ──────────────── 43. This principle is particularly worth learning for entrepreneurs. Many CEOs ask: "How does Google hold meetings?" "How does Amazon write memos?" "How does Netflix do performance?" Then: copy it. Wrong. Because: Best Practice is Context-Dependent. Amazon's six-page Memo is effective for Amazon. It does not mean: your 12-person startup should also write six pages. ──────────────── 44. Excellent organizations do not adopt "best processes" but rather: design processes that suit their own talent structure. The early Sean Parker of Founders Fund: might not follow the traditional fund way of "sitting in the office". But occasionally appears: Facebook. Spotify. The economic value of this one-time judgment: may exceed ten years of work by a diligent ordinary investor. This is the very special aspect of Power Law companies. ──────────────── 45. VCs should not measure people by "working hours." Assume: GP A: looks at: 1000 companies in a year. Writes: 200 memos. Makes: 50 investments. No super winners. GP B: looks at: 100 companies in a year. Makes: 5 investments. One of them: $20B outcome. Who creates value? B. So VC is a very typical: Output ≠ Activity. ──────────────── 46. This is also why Founders Fund's variable carry is very interesting. According to Singerman's description: a partner's carry in the current fund, will be significantly influenced by the actual investment return contributions from the past one to two funds. Carry: can rise. It can also fall. This is not the precise algorithm in public legal documents, but rather Singerman's description of the internal mechanism. ──────────────── 47. This mechanism solves a very tricky problem in traditional VC. The biggest advantage of Equal Partnership: team collaboration. No project disputes. No personal empires. But the downside is: If a Partner does not generate returns for a long time, what to do? Everyone has the same carry. If it cannot be reduced: the only option may be: Fire. ──────────────── 48. Variable Carry provides a middle state. Not: 100%. Or: 0%. But: Economic Gradient. Contribution decreases: carry decreases. But can still: mentor, network, help the portfolio, do some valuable work. This reduces: Binary Politics. ──────────────── 49. But this mechanism also has very large side effects. If not designed well: every GP may start: grabbing attribution. This is "my deal." This founder was discovered by me. This follow-on should count as mine. Internal politics may explode instead. So variable carry can only operate when: Attribution Rules + Trust are both very strong. ──────────────── 50. This is why Benchmark's equal partnership can also be the correct answer. Singerman himself did not say: Benchmark is wrong. On the contrary, he acknowledges that Benchmark has taken the egalitarian model to the extreme. So what should be learned here is: There is no universally optimal partnership model. Only: A model that fits the characteristics of your team. ──────────────── 51. GPx is essentially abstracting Singerman's ideas one layer further. Previously: Brian looked for: Spiky Founders. Now: Brian looks for: Spiky Investors. Previously: Helping founders to double down. Now: Helping GPs: When they truly discover a super winner, to have enough money to bet. ──────────────── 52. This solves a significant structural problem for Emerging Managers. Assume a new fund: $50M. A certain seed project: Initial investment: $2M. A few years later: The company proves to be very strong. Now Series B: Needs to invest again: $10M. Here comes the problem. $10M: Equals 20% of the fund. The fund manager might judge: This is the best company of his career. But: He has no money. ──────────────── 53. Thus, traditional Emerging GPs have three choices: First: Sell ownership. Second: Open an SPV. Third: Bring in a large fund. Result: When a truly massive winner appears, because the fund size is too small: Ownership gets diluted. This is a very cruel structural problem in VC. ──────────────── 54. What GPx wants to do, in essence, is: Add balance sheet to conviction. You are responsible for: Finding companies. Building founder trust. Making early investments. If you later dare to bet: 20% of your fund, only then does GPx believe: This is a high-quality conviction signal. Then: It provides larger programmatic capital to help you continue participating or even preempt subsequent rounds. ──────────────── 55. Note, "20%" is not an ordinary follow-on. This is a very nice: Revealed Preference. Investors say: "I am very optimistic." It means nothing. Ask: How much are you putting in? That matters. In economics: Actions are more real than words. ──────────────── 56. If a GP: $100M Fund. Is willing to put: $20M into one company. If it fails: It will severely impact the entire fund. So this is: Skin in the Game. GPx uses concentration as: Signal Extraction Mechanism. This design is very clever. ──────────────── 57. It's a bit like Poker. A player says: "I have a great hand." No information. If he: Goes all-in. Suddenly there is information. Of course: He might also bluff. But the cost is high. Similarly: A GP says: "This is the greatest company I've ever seen." Useless. If he puts 20% of the fund in: The information suddenly becomes significant. ──────────────── 58. What GPx is really betting on is the "two-layer choice." Traditional VC: LP ↓ GP ↓ Startup. GPx: LP ↓ GPx ↓ Elite Emerging GP ↓ Breakout Company. It seems to add a layer. But the goal is: To use a layer of special "manager selection": To enhance final portfolio access. ──────────────── 59. This is different from a typical Fund-of-Funds. Typical FoF mainly does: Diversification. Invest: In 20 funds. GPx is more like: Manager Incubator + Co-Investment Engine. Public reports show it plans to allocate about 20% of capital to emerging-manager funds, with the remaining capital used more for co-investing with these managers in breakout companies. This is the real innovation of GPx. ──────────────── 60. I suggest temporarily removing the point in your materials about "not charging GP carry at all." Currently, publicly reliable information confirms that: GPx: Invests in emerging managers, Provides advice, Helps connect LPs, And provides programmatic follow-on capital. However, the specific terms regarding all management fees, carry, and co-invest economics between GPx and underlying managers have not been made public. Moreover, TechCrunch even pointed out that the FoF model typically faces: dual-layer fee issues. So do not write it as: "There is absolutely no dual-layer carry." Unless Singerman later publicly provides the contractual mechanism. ──────────────── 61. The value of GPx's "Consigliere" may actually be more important than money. The most important quality of a typical LP is: "Not causing trouble." But when a new GP encounters: The biggest portfolio company having problems, Partner conflicts, LP fundraising, Founder crises, Whether to sell, Whether to concentrate, Who to ask? Very few have actually managed funds that returned billions of dollars. Singerman and Linden want to incorporate: Institutional Memory as part of the product. ──────────────── 62. This is actually creating the VC industry's own "Y Combinator." YC does: Founder Network. GPx hopes to create: GP Network. Find: Excellent early investors. Provide: Capital. Provide: Experience. Provide: LP access. Then: When he finds a breakout, continue to support. If successful: What GPx holds is not just an edge of one startup. But: A Network of Edges. ──────────────── 63. This is why GPx has great potential value. A VC themselves: Can seriously look at a limited number of projects in a year. But if GPx connects: 20 top niche GPs. Each has: Completely different networks. AI. Defense. Biotech. Consumer. Crypto. Industrial. GPx equals building: A Distributed Sourcing Network. This is essentially a: Human Information Network. ──────────────── 64. Moreover, it aligns very well with Singerman's "spiky philosophy." GPx does not want these managers to become: The same kind of GP. On the contrary: The more different, the better. The premise is: Everyone has their own: 15/10. This combination is ultimately similar to Anduril: Not looking for five "standard founders." But: Looking for five completely different super abilities. ──────────────── 65. This could actually be called: Portfolio of Human Monopolies. Everyone should have: A localized monopoly ability that is hard for others to replicate. This is worth much more than: "Everyone is smart." ──────────────── 66. This also explains why Singerman believes solo GPs have a special advantage. When a solo GP puts: 20% of the fund into one company: There is no: "It's all Partner B's fault." No place to hide. As a result: It is entirely yours. This creates: Extreme Accountability. In a large partnership, Judgments can sometimes be diluted by: committee. ──────────────── 67. But solo GPs also have serious problems. No one challenges your judgment. No partner succession. Key-man risk is extremely high. Bus factor = 1. One person's network is also a ceiling. So solo GPs are not: Inherently superior. They are just particularly clear on: Conviction Attribution. ──────────────── 68. When Singerman says "SPV is dead," it should not be taken literally. SPVs: Are certainly not dead. There are still plenty of SPVs in the market. What he means is: Top founders are becoming increasingly sensitive to cap table quality. Good companies do not lack capital at all. Why accept: Dozens of opportunistic SPVs? Founders prefer to choose: Long-term capital, People they know, And those who can truly help the company. ──────────────── 69. This represents a shift in VC from "Capital Scarcity" to "Trust Scarcity." 20 years ago: Entrepreneurs lacked money. Today, top AI founders: May receive: Dozens of billions of dollars in capital intentions in a day. So capital itself: Is a commodity. What is truly scarce is: Founder Trust. Singerman clearly stated in the interview that in an environment of extreme capital abundance, trust may even surpass brand. ──────────────── Seventy, this statement is very important for the entire future VC industry. If capital becomes a commodity, the traditional VC moat: money declines. What remains: Access Reputation Judgment Founder Relationship Company-Building Ability. So in the future, VCs will increasingly resemble: Talent Agency + Capital + Strategic Partner. ──────────────── Seventy-one, Airbnb being ranked 30th and still being able to enter is the old world of Brand + Capital. But the best founders in the future: may increasingly emphasize: Who really helped me? Who accompanied me? Who introduced core employees? Who supported me in my worst times? This is called: Relationship Capital. It cannot be raised temporarily. ──────────────── Seventy-two, this is also why the truly highest asset for VCs is not Fund Size. SoftBank can have more money. Sequoia, Founders Fund, Benchmark are truly powerful because: some of the world's best founders: call them before fundraising. This is called: Access Alpha. Without access: no matter how strong the judgment, it is useless. ──────────────── Seventy-three, speaking of AI: Singerman's PvP/PvE classification is very interesting. PvE Humans facing problems. For example: certain mathematical proofs. Once AI: completely solves it. The significance of humans continuing to solve it "for practical value" may decline. ──────────────── PvP Human vs. Human. Football. Chess. E-sports. Entrepreneurship. Investment. Even if machines: are a thousand times stronger than humans, humans still want to watch: competition between people. ──────────────── Seventy-four, chess has already proven this point. Deep Blue defeated Kasparov in 1997. Today, chess engines are much stronger than top human players. The result: Magnus Carlsen's matches: did no one watch them? On the contrary. Humans still watch. Why? Because what we consume is not: Optimal Move. What we consume is: Human Drama. ──────────────── Seventy-five, sports are even more obvious. Cars run faster than Usain Bolt. So: Is the 100-meter race meaningless? Not at all. The audience is not there to see: the absolute fastest way to move on Earth. But to see: the limits of humanity. ──────────────── Seventy-six, this provides a very important judgment framework for AI entrepreneurship. Ask: Is this product solving: Outcome Problem or: Human Experience Problem? If consumers only want results: AI can easily replace it. For example: Translating a contract. Summarizing documents. Generating basic code. ──────────────── If the value itself comes from: human experience, identity, competition, social interaction, scarcity, AI replacement becomes much more difficult. For example: Sports. Real art. Dating. Community. Competitive gaming. Live performances. ──────────────── Seventy-seven, this is why "AI can do it" does not equal "humans no longer do it." AI can write songs. Humans will still sing. AI can play chess. Humans will still play chess. AI can simulate football. Humans will still play football. Because: Utility ≠ Meaning. AI may consume Utility. But: Meaning Economy may still be very large. ──────────────── Seventy-eight, this may even make "Human-made" a new scarce asset. When AI content is generated infinitely: if something is clearly: made by real people, competed by real people, performed by real people, written by real people, it may instead generate: Scarcity Premium. This is like: in the era of machine production, Handmade can sell for a higher price. ──────────────── Seventy-nine, so one of the biggest long-term paradoxes of AI may be: The stronger AI becomes, the more valuable human authenticity is. Not in all fields. But: Art, Entertainment, Competition, Identity, Luxury goods are likely to be so. ──────────────── Eighty, returning to Singerman, the highest level of the entire interview is actually not "how to find good founders" but rather: Do not optimize for the average. The school system likes: average. Large companies in hiring like: average. Traditional performance systems: average. But VC returns: are completely not average. Startup outcomes: are completely not average. Talent contributions: are also completely not average. So a system designed for a Power Law world should not adopt the scoring rules of a: Gaussian World. ──────────────── Eighty-one, this is a very deep statistical problem. In an ordinary world: height. blood pressure. exam scores. Many variables approach: Normal Distribution. The average is very important. But: Entrepreneurial returns, wealth, network traffic, VC outcomes are closer to: Power-Law Distribution. A few tail results determine the overall outcome. ──────────────── Eighty-two, therefore management methods must also change. In a Normal Distribution world: reduce failures. increase averages. standardize. In a Power Law world: look for extreme right tails. tolerate the strange. tolerate the imbalance. tolerate failures. heavily invest in a few exceptional results. This is the true mathematical foundation of the Founders Fund and Singerman philosophy. ──────────────── Eighty-three, so "Spike" is actually not about personality traits but rather: Portfolio Construction applied to humans. You do not want everyone: to be good at everything. But you want the whole team: to have all the important capabilities. And each key capability: is best handled by world-class individuals. This is a completely different organizational philosophy. ──────────────── Eighty-four, this is extremely important for entrepreneurs. If you are: a world-class Product Founder. Do not look for: another Product Founder like you as the second partner. Look for: world-class Distribution. The third: Operations. The fourth: Technical Infrastructure. This is called: Complementary Founder Architecture. ──────────────── Eighty-five, and the "common ground" among co-founders should focus on Values, not Skills. Skills: should be different. Values: must overlap. For example, the founding team of Anduril has vastly different abilities, but they share the belief that: The U.S. defense technology system needs to be rebuilt. Brian Schimpf himself publicly recalls that several former Palantir colleagues long discussed establishing the next generation of defense companies, which later combined with Palmer's vision. This is: Shared Mission + Complementary Skills. Very strong. ──────────────── Eighty-six, what happens if only skills are shared, and the mission is not? Everyone wants to: control the product. compete for CEO. compete for credit. And then they part ways. If only the mission is shared, and skills are the same: the team has a serious capability gap. So: a great founder team can be described as: Values Alignment × Skill Diversity × Mutual Respect. ──────────────── Eighty-seven, from the perspective of capital allocators, there are six things I want to learn from Singerman. First, do not use average scores to identify people. Find 15/10. Second, know what your game is. Others' successful methods may not suit you. Third, acknowledge that others saw it right first. Airbnb being 30th can still win. Fourth, position sizing itself is a judgment. Without position sizing, conviction is just a slogan. Fifth, concentration is not to appear brave. Only super assets deserve super positions. Sixth, capital will ultimately become commoditized, but Trust will not. These six points actually apply to: investment, entrepreneurship, recruitment, career development. ──────────────── Eighty-eight, for personal career development, the "Spike Philosophy" is also very practical. The vast majority of people ask: "How do I compensate for my shortcomings?" Of course: fatal shortcomings must be compensated. But the more important question is actually: What can I do to be in the top 1% globally? If the answer is: None. Long-term competition will be very difficult. Because AI will further lower the value of: "average is good." ──────────────── Eighty-nine, the AI era rewards Spiky People even more. Because many ordinary abilities: writing ordinary code, writing ordinary articles, doing ordinary analysis, ordinary design, ordinary research are becoming cheaper. What remains truly valuable is: Taste Judgment Network Deep Domain Knowledge Leadership Extreme Creativity Trust. In other words: AI will compress the Average and amplify the Spike. ──────────────── Ninety, this is also why the most dangerous talent strategy in the future is: "Cultivating all-round employees." No. AI itself is becoming: A universal assistant. What people need to provide is: Non-Commodity Edge. ──────────────── Ninety-one, GPx is essentially this bet It believes: The best outcomes for future VCs may not all come from: Giant platform funds. There will also emerge: Some strange, Highly specialized, Strong personal brands, Deep networks: Elite Emerging Managers. GPx hopes to identify them while they are still small: This is: VC of VCs. ──────────────── Ninety-two, so if GPx succeeds, there will be a very beautiful second-order Power Law First layer: A few startups Generate most of the returns. Second layer: A few GPs Generate most of the venture returns. GPx attempts to first solve: Which GP owns access to the next right tail? Then through: Programmatic capital Expand that right tail. This is a very advanced capital structure design. ──────────────── Ninety-three, its biggest risks are also very obvious First: Manager Selection may also be extremely difficult. Second: The past Spike of the GP may not be sustainable. Third: 20% concentration may also be a false signal. Fourth: Multi-layer fund structures will increase costs and structural complexity. Fifth: The best founders may not allow GPx to subsequently enter the cap table. So GPx itself has not solved the core problem of VC: Selection. It has only created a different: Selection + scaling mechanism. ──────────────── Ninety-four, this is also why no investment model should be deified The Sequoia model can win. The Benchmark model can win. The Founders Fund model can win. YC's extensive seeding can also win. Solo GPs can also win. The most important thing is not: Which strategy sounds smartest? But: Which strategy fits your edge? This is the true central idea of Singerman's entire interview. ──────────────── Ninety-five, so I would compress his philosophy into a formula Edge × Self-awareness × Position Size × Trust Edge What do you have that others don’t? Self-awareness Do you really know your strengths and weaknesses? Position Size When opportunities arise, how much are you willing to bet? Trust Why would the best opportunities choose you? Only if all four are true: Can there be: Outlier Return. ──────────────── Ninety-six, and this formula applies not only to VC For startups: Product Edge × Founder Awareness × Resource Concentration × Customer Trust. For career development: Skill Edge × Self-awareness × Focus × Reputation. For investment: Informational Edge × Calibration × Position Sizing × Discipline. The essence is: Do not evenly distribute limited resources. ──────────────── Finally, I think the most memorable takeaway from this Brian Singerman interview is not the "spiky founders" But this sentence: Top competition is not about striving to achieve 80 points on all dimensions, but discovering a dimension where you can achieve 150 points, which is hard for others to catch up with, and then reconstructing the entire game to let this dimension determine the outcome. The strength of Founders Fund back then: Was not having the world’s best standard processes. On the contrary. Sean Parker has his own game. Peter Thiel has his own game. Brian Singerman has his own game. They did not try to: Make everyone the same. But rather let: Different extreme advantages generate compounding within the same organization. This also explains why the case of Anduril is so beautiful: Palmer does not need to become Brian Schimpf. Brian Schimpf does not need to become Palmer. Matt Grimm does not need to become Trae Stephens. A truly great team is not: Five equally excellent people. But rather: Five who are exceptionally excellent in completely different directions and are willing to work together for a common mission, the "spikes." If we compress it further from a capital perspective: The real art of venture capital is not "often judging correctly," but recognizing, acknowledging, capturing the outliers when they truly appear, and ultimately putting enough capital into them. The 30th investor in Airbnb can make the second most money; Anduril has gone from a $17.5 million Seed to a current confirmed valuation of $61 billion, and at one point explored $100 billion level financing; These cases collectively demonstrate: In a Power Law world, discovering the truth a little late is not scary; what is truly expensive is not daring to heavily invest after discovering it. This interview was released just two days ago, and GPx, Anduril's new financing, and changes in Founders Fund's investments are still very fast. We can continue to track which emerging GPs Singerman actually chooses next—this will be a better test of this "spiky investor" theory than what he says.