Sarah Guo sees four to six new companies a week. In her first couple of months at her old firm she saw 500.
The common worry among investors in this market is moving too slowly and missing the companies that matter. Guo's runs the other way: that large research bets are being funded on the strength of who the founder is, because nobody in the room can evaluate the science.
"I do not want that future very clearly and I don't think we are going to end up there."
Guo founded Conviction in 2022, wrote the first check into the drug-discovery company Chai Discovery, backed Sunday Robotics in the first meeting, and is close enough to the frontier that this show published a profile of her in July under the title Sarah's Wager.
I listened to the full interview so you can skip it. 60 minutes of audio, 27 minutes of reading.
Here are the 19 takeaways that matter.
👤 Guest: Sarah Guo, founder and managing partner of Conviction, the venture firm she built in 2022 to back AI-native companies from their earliest days, and a former general partner at Greylock
🎙️ Host: Patrick O'Shaughnessy, founder and chief executive of Colossus, the media company that publishes this show
📰 Published: 1 September 2026 on YouTube (Invest Like the Best with Patrick O'Shaughnessy)
🔴 YouTube | 🟣 Apple Podcasts | 🔗 Show notes | ⏱️ 1 hr | ✅ Time saved: 33 min
Key Takeaways
Guo does not want a future in which the owners of one to three frontier models take the economy, and does not expect one
The early-stage field she entered was less competitive than it looked, and her edge was focus rather than insight Her partner Mike tracks a group of about 250 people at the frontier that the firm tries to know personally
Law was the first application she backed, because legal work is language with a structure
Researchers at the big labs feel less decisive as compute spend and headcount rise
The constraint on building AI energy is regulatory and social, not technical The actual task is convincing the people of New York that they want a data center
Investors are substituting a founder's pedigree for a view on the business, and she thinks that is dangerous
A robotics company less than two years old expects semi-humanoid robots doing tasks in homes, in beta, this year
She sees 4–6 new companies a week, against 500 in her first couple of months at her old firm She raised her first fund on a two-page background document and an admission that she did not yet know what the firm would be
Restricting open-source models inside the United States would bind only law-abiding American businesses
Software can make money serving pharma, which the industry did not believe What moved her was a $10M contract, not a scientific result
She will not spread an investment decision across a partnership Somebody has to own the decision, or she does not know how the business runs
1. Individuals change outcomes
O'Shaughnessy opened by asking what this moment feels like from inside it. Guo said the honest problem for an investor is that there is nothing to test a decision against.
She said she cannot slow down even when she wants to, and borrowed a line from an investor friend she had spoken to the night before. "I keep saying I want to press the brakes as hard as I can but I'm not doing it I'm going 90 miles an hour." The question underneath it is whether you miss the opportunity, or repeat the mistake of every technology boom and bust before this one
Colossus had published a profile arguing she is making a specific bet against the largest AI companies, and O'Shaughnessy asked whether that was right. Guo answered with a different frame: "I believe in like the great man and great woman theories of history." Her claim is that high-agency people with the right risk capital change outcomes. Whether a competitive Western open-source model exists at all comes down to whether anyone raised the money, gathered the talent and built the infrastructure
She rejected the idea that this is a war. She said she works closely with the big labs, co-invests with them and has friends inside them
What she rejects is the concentrated end state. "I do not want that future very clearly and I don't think we are going to end up there."
Asked whether she wants to be a great woman in that sense, she said no, and not out of humility. "It was just like it's not in my set of goals, right?" "Like I want to be the best investor in the things that I try to do." She had expected to be a software entrepreneur. She chose investing because she is curious, likes being right, and is motivated by working with extraordinary people and making them more successful
2. The early bet was simple
Guo said the early-stage field was less competitive than it looked when she started, because firms were growing and generational transitions were under way at the same time as a large technology shift. "So, all you have to do is like take the risk and be focused and then it's an execution play." She said the hard part is effort rather than cleverness, and that what matters now is the bar she sets for the people she works with She and her partner Mike began with a set of existing relationships
O'Shaughnessy pushed back that outworking everyone does not explain the result, and put two pieces of evidence to her. The first was a group of people the firm tries to stay close to: "Your partner Mike said something interesting to me a couple weeks ago, which was there's sort of 250ish people that he thinks about or you guys think about" He described them as "Just like the people that are actually like showing up in the morning and pushing this whole thing forward." — entrepreneurs and researchers doing the most interesting work at the frontier
The second was a limited-partner survey, and it is the host's claim rather than the guest's. "I won't I won't name them, but there was there's one well-known LP that does this like survey every year of what all the other fancy LPs who they most want to invest with." "And you were either number one or two."
3. Law is text in, text out
Guo said the firm went looking for professions whose work already fit what the models could do, which is the reverse of the standard advice. "Lots of people be like this is nonsense, right?" The orthodoxy is to work back from the customer's problem. She said she wants to do both
Law fit because the work is text going in and text coming out. "And if you look at Harvey and the function of the law, like rationally, if you think that we can do next token prediction with language and you knew that in late 2022, then law is structured language." She is not a lawyer. Her reasoning was that the job involves reading a lot of documents, that retrieval existed, that text has to be generated, and that firms and common law supply an enormous body of precedent
What decided it was how far Harvey's founders were willing to project. From answering a question about a California landlord-tenant agreement, which she called somewhat trivial, they were talking about "I can project to doing a Activision Blizzard M&A and doing 85% of the work." The founders, whom she named as Winston and Gabe, believed AI would transform the practice of law at a point when saying so out loud was unusual She said the appeal was the ambition of what was possible plus the technical logic of why it would work, and that this was a different decision framework from the one others were using
4. Researchers feel less needed
Asked what the frontier group is talking about now, Guo led with the competition. "It is a violently competitive landscape." She said that was true twelve months ago and is more true than it was two or three years ago, that it is now global, and that the global part carries an insecurity because it breaks the story people had been telling
A belief roughly a year old among researchers is that models which improve models put exponential intelligence one or two years away. She set the belief out and then questioned it, quoting Andrej Karpathy against himself: "I've thought it was two years away for about 10 years." Karpathy believes it again, she said, and added that nobody can say
The change she thinks matters more is scale, and she put it as the question a researcher now faces. "I need 750 billion dollars of compute spend and and we have many thousands of people working on this problem like I think people feel less ownership of the of the outcome." When a lab was 200 people, she said, how it got there was up to every single person. Two beliefs are circulating now, and both take agency away: that the model will do the work anyway, or that only compute scale counts
O'Shaughnessy asked what a top researcher does if both are true. Guo said some scientists want to work on the problem whether or not their presence changes it, and answered the same question about herself A founder had asked her which of the companies she backed would not have been funded without her. Her estimate was "couple of them 5% 10% max" Resourceful founders find other investors, she said, but they might not have found the next hundred million dollars of compute
5. Energy needs public consent
Compute is the first thing she said worries her, and the timeline people now plan against runs past 2030. She said an infrastructure leader at one of the hyperscalers had told her that week that "there was nothing that is going to move the needle for us at sufficient scale before 2030" Her reaction: "I was like oh that's depressing." The binding question, she said, is where sufficient natural gas comes from
She does not think the constraint is technical or financial. "I don't think it is a technology or capability or capitalism problem. I think it is a regulatory problem and an alignment problem." By alignment she said she does not mean AI alignment. She means getting people to agree
The example she gave is siting. "If you want to build data centers in New York, you need to convince the people of New York they should want data centers there or America should want data centers there." On nuclear power as base load, she said the price only falls if enough is built, and enough is only built if people are convinced it is safe. The cost curve itself is understood
The physical supply chain is the harder half and has no shortcut. She said learning to build things, the tacit knowledge, the labor and the raw materials cannot move at the speed of software, and that the only way through it is to invest in it
6. Pedigree is doing the work
Guo said research founders now have to make a capital-intensive bet legible to investors who cannot check it. "I'm not a research scientist, you're not a research scientist and all of the capital is not research scientists." She had told several of her own founders that the quality of their storytelling is obviously important to their success
Her worry is what fills the gap when the investor cannot evaluate the science. "There's a lot of proxying of judgment to pedigree or to other legible signals" References and referral sources are useful, she said, but decision-making built on them is less fundamental
She described a debate with an investor she rates highly. She asked him to explain what the company would be that would be big, and his answer was about the founder: "I've known the quality of the person for eight years but the business like the technical theory in the business don't make sense to me." Her conclusion is that a large research bet with no view on the business, only on the person, is dangerous, and that she may not be any better at deciding but wants the intuition first
7. Robots need cheap data
Asked which researcher had most impressed her, Guo named the two founders of Sunday Robotics, Tony Zhao and Cheng Chi. She and her partner Pranav met them as Stanford PhD students who had already worked at Toyota Research, DeepMind and Tesla; she put them at about 25, and thinks one of them did not finish the doctorate She said it took her a while to get oriented in their body of work, and that they have contributed most of the interesting ideas in robotics AI over the past four years
The problem they went at is where robot training data comes from. "It is believed in robotics that if we just had the internet of robotics data like yeah we'd have like fully general robots everywhere." Their work is on collecting that data as cheaply as possible while still covering the spread of real environments and tasks, and on how the shape of the collection interacts with what the model learns
What she says is unusual is the pace from research to shipped product. The company is just under two years old, and the team expects semi-humanoid robots doing tasks in people's homes, first in beta, by the end of this year Her caveat: "nothing is true until it is shipped" She said the speed is mindboggling, that most people in robotics now treat home robots as a question of when rather than if, and that even so the team's timeline surprised her
8. An 8 or 9 on day one
Guo said she is instinctive about people and usually knows immediately. The firm rates companies on a 1 to 10 scale, and she said she is often at an eight or a nine on first contact when she already knows the founder's work
The work between the instinct and the decision is looking for what she has missed. "And what I'm then doing between that and a real decision is often figuring out like what are the holes in my understanding where my judgment of like their premise or them is incomplete or wrong, right?" The firm invests across biology, defense, robotics and law, so she cannot know every domain, and spends the next day to few weeks working out what everybody else believes about a space
She writes a memo, and did so when the firm was only her. "I'm a memo person." She used to send the memo to a trusted investor outside the fund — Dylan Field was one — for a second read, and now takes it to her partners
The one exception is a founder good enough that the idea would not have to make sense to her. She said if Brett Taylor wanted to dig in volcanoes or do dog streaming she would back him, then noted he would not do that: "So, so there is some class there is some version of like I don't even care if it doesn't compile in my brain. The person is so undeniably good that I would just back them." She then argued against her own exception, because judgment is the thing she is buying and she cannot assess judgment on a plan she does not understand: "And so, can you really be an eight or a nine if you don't get it? Like, no." She had met an interesting company that week with her partner Bella and was positive on instinct but did not yet know enough about the science
9. 4–6 new companies a week
Two thirds of her time goes to companies the firm already owns. "I think I spend 2/3s of my time working on portfolio company stuff." That covers recruiting, helping founders think through problems and trying to influence the outside ecosystem. Then fundraising. Looking at new companies is the next largest piece
The volume of new companies is deliberately low, and she said she gets paranoid about it. "I probably see four to six new companies a week it's not a very high volume my first couple months at my old firm I saw 500 companies" She said she now has more calibration and more confidence in what she can tell
The balance goes to learning from people outside her own asset class. She said she had never spent much time with public-markets investors and finds how they think educational, and that she was spending time with a pharma company on what AI does to its business Her argument is structural: an early-stage investor inside a large firm can be myopic because that firm's own ecosystem is big enough, and a small firm has to be ecosystem-oriented She also leaves calendar room for curiosity, and does work she described as bridges to Washington and external communication
10. She raised on an admission
Guo said she refused to construct a differentiated story for the fundraise, and gave prospective backers the same advice she gives founders. "I stand with the belief that I advise entrepreneurs with you should understand people's objections to what you are doing and their questions but you should not tell them what they want to hear." A private-equity friend had told her every part of the funnel needed a specific claim about what the firm would do
What she offered instead was an admission. "And I'm like, let's be honest, I don't know yet, but I need to like raise some money so I can go like experiment and figure it out." She gave a two-page document on her background and investing history, claimed she was good at identifying extraordinary people and at supporting them, and said the rest was firm culture plus execution
Some limited partners did not like it, and she said she understands why. The ones who backed her had to take an investment-committee memo that amounted to "She's going to execute like hell. We'll find out."
She will not advise other managers to copy it. Outcomes differ, she said, but telling people what you actually believe makes life simpler, and being convinced the world works differently than she thought is the thing that excites her most as an investor
She said she admires investment managers who treat the firm itself as something to build. She named Josh at Thrive among them, and said what she likes is the encouragement that you can do new things and express your opinions through the shape of your firm
11. She was raised not to care
What she took from two entrepreneur parents was a rule about other people's opinions, and she said they treated it as a moral question. "They were like you cannot ever worry about what other people think." She said there was no moment of her childhood when she felt they were not there for her even though both worked all the time, and that a family can make a child feel like the center of its world while still being whole people with other interests She listed the family values as integrity, thinking for yourself, focus and team spirit, and said independence of thought is the least generic of them
She admits to one thing she does worry about. "I worry about raising people's competitive hackles in the ecosystem." The reason: "Because I'm a friendly person. I want to be friends with everybody,"
What she objects to is the ownership stance of the traditional Series A and B firm, not the competition. "I'm going to own 18 to 25% of this company and take the board and you're going to own none of it is like is not conducive to like a lot of collaboration." She contrasted it with public-market investors, who she said love telling you their best ideas so that others buy in behind them, and said that orientation appeals to her because she is positive-sum The example she reached for of something she would like to talk about openly is gaming and entertainment, which she said is going to be totally different
12. Cynicism is not realism
Asked who inspires her, Guo started with watching her parents build a company. "I saw my parents build a company. I was like, this is so cool, right? Like it's us against the man and like the man is very big companies." What she takes from entrepreneurs is that you can make something out of nothing because you can see a better future, and do it quickly
What she called poisonous is the belief that the whole game is marketing and network. "But when folks are very cynical about how the world works, how entrepreneurship works, like that it's just nepotism and like Twitter is useful. I think that's nonsense, right?" She is careful to say nobody denies that brand and network are real
Her counter-claim is that the plain version works more often than people expect. "I'm like, if you focus on value and treating people well and you work with extraordinary people and the vision is worthwhile, like that works more times than you'd think." She named Tuhin as someone she finds inspiring for exactly this, summarizing his position as doing right by the customer and winning, and said it seems to be working
13. A ban binds only US firms
O'Shaughnessy framed the policy question: open-source models that are not American are competitive at the frontier and appear to be built at least partly on American work. He asked what should happen, what will happen and what it means for business
Guo separated what she thinks is healthy from what has already occurred. "So I'd say like the cat is out of the bag and these are in use everywhere." Three years of increasingly competitive open models, largely from China but also from the United States and Europe
Her economic case does not require any view of the labs. "There are a huge number of instances where it is too expensive, too sensitive or too slow to use the like model from the frontier providers today." She expects that set to grow rather than shrink as more is attempted with AI, because doing more is expensive
She said a domestic restriction would bind only the people who obey it. Attackers and adversarial users are unaffected, so the effect is to slow American businesses down or move profits between pockets: "So, you're just you're restricting your own people."
Her alternative to speculation is testing. "People are very worried about backdoor like behaviors in Chinese models." "My view would be like there should be testing and understanding of these models at the frontier." She does not dismiss the safety question. Models capable of defensive cybersecurity and biology work are capable of the offensive versions, and she said that reality has to be looked at directly
The end state she expects is intelligence cheap enough that access stops being the constraint. Her phrase for it was "broad access to intelligence too cheap to meter", which she credited to Sam. "Businesses want it to control their own destiny for economics for capacity." Her argument for why diffusion needs an ecosystem rather than a lab: individuals have use cases no frontier researcher will imagine, and models still have to be got there
14. Cheap AI is not automatic
O'Shaughnessy asked whether abundant cheap intelligence could fail to arrive, the way abundant cheap nuclear power did. Guo said it could, and could also arrive without the United States being competitive in it: "Yes, absolutely. And I also think I can imagine very easily a world where we don't have that in a competitive way because it is essential to economic competitiveness and national security."
Her first premise is that the American industrial base cannot be rebuilt without automation. If people are not imported, domestic labor is expensive and some skills are missing, then producing far more goods on a more resilient supply chain does not add up any other way The wage arithmetic she used: "people in the United States do not want to work and should not want to work for $13 an hour doing a very inhuman job"
The risk she names is political rather than technical. People may rationally fear what AI does to jobs, dislike rent captured by a small number of technology firms and reject the idea of a permanent underclass, and connect that to an anti-capitalist position That contingent, she said, can slow the build-out of energy, infrastructure and industrial capacity, and compute is one of the most important inputs "I think we're going to start talking much more about compute independence."
15. Every input needs a backup
Asked what compute independence actually means, Guo went upstream of the chip. "There's particular kinds of glass, very, very important that's controlled by basically one company that TSMC has like a monopoly on the supply of." She said she had her TSMC mug with her
The shape of the problem, she said, is energy independence. Work backwards from a data center full of GPUs — cooling, powering, training and inference — and every input has a global supply chain, parts of which run through places that are not necessarily stable or accessible to the United States and its allies
The goal is not to make everything domestically. "But having more than one source is a position that everybody wants to be in." Comparative advantage is real, she said. She pointed to work she attributed to Jacob Helberg going part by part through the supply chain to ask where more capacity and independent paths could be built
What the firm has actually funded. "When we think about the components that we've invested in, like we've invested in the labor gap for data centers and robotics, we've invested in nuclear energy, we've invested in alternative chip architectures." They keep looking at data-center builders and solar and battery installers and have not invested. She said financing is probably the dominant factor in those businesses, and she is a technology investor who wants to understand the durable product
The internal debate this settled was semiconductors. She said venture investing in semiconductor companies was a bad business for a long time, and that her partner Bella's work on the sector changed the firm's view The reason is demand: buyers are consolidated and at scale, and they want supply-chain independence themselves, which changes the risk equation for a venture-backed chip company Her way of putting the buyers' motivation is that everyone cannot be stuck on one production line at TSMC She said the firm starts from aligned beliefs about direction and then argues about whether a market is friendly to a venture-backed company at all — space, solar, batteries, nuclear, turbine manufacturing, robotics and biology each get their own version of that argument
16. You can sell AI to pharma
Biology is the debate she says the evidence has already settled for her. "You can create and capture enormous value with models in biology."
The conventional wisdom was that serving pharma does not pay unless you make the drug. She described the traditional structure: firms find principal investigators, own about 40% of the company, assemble candidates and take them some distance down the risk path, and most of it does not work Her shorthand for the received view is that software investors were told you cannot make money selling software to pharma or build a platform business there
Her counter-example is the first check the firm wrote into a company. "We're the first check in a company called Chai Discovery. And Chai is working with a number of top 10 pharma companies in really significant ways to accelerate some part of the R&D process." She had been looking at computational biology companies for five years or more at that point The evidence that moved her was commercial rather than scientific: "There's no genius here. I'm like that's a $10 million contract you know." She also said you talk to the scientist at the customer, or to users of the tool, because the customer knows whether it is valuable
What she is waiting for is a visible case. "I think the like light bulb moment for the industry is when we will have a new indication or a new drug that like clearly the trajectory of the thing was changed created by AI" She expects a wave of investment behind it, and said she is impressed by how fast pharma and healthcare have accepted that this changes their business She still flags regulation, the speed of the physical world and safety as things models do not overcome
17. One person owns each bet
Asked why the firm is called Conviction, Guo said the name is aspirational rather than descriptive. "It's aspirational, right?" The most traditional form of early-stage investing, she said, is to start early, take a significant position, never sell it and work on the company until it works
Both she and her partner Mike had backed companies that took years to become obvious. She named Figma, Notion and Rippling, and said Baseten's first couple of years were also non-obvious Nobody bets on a company hoping it takes four or five years to find the thing, she said, but everybody is a product of their own investing experience
The thing she will not do is spread the decision across the partnership. A friend who runs another firm told her "We don't have individual ownership of our investments." and her reaction was that she could not run a business that way "Because somebody has to own the decision" Her version is that every input goes to one person, who has to be convinced and then has to decide
18. The edge is holding a view
She said she has no instinct to be contrarian, unlike her friends at Founders Fund, and that the discipline is something else. "I think you like just need to find the truth." What she guards against is the dominant narrative of the period and the declarations of important players in the ecosystem
The link between truth and return is pricing. "You want asymmetric information and then the confidence to like hold the opinion when other people haven't come around to it yet." She said she spends a lot of thought on how to have better information than others and then how to protect the firm from noise
Where the information comes from is who she spends time with: portfolio founders, founders outside the portfolio doing something that surprises her, and smart people who believe something she does not
The miss she volunteered is Suno. A mutual friend asked her to invest and she declined, because she doubted how many people want to make music and how much consumption there would be. "I knew Mikey, a mutual friend of ours who was an investor, like asked me to like invest and I stupidly said no." What she says she got wrong is broader than one company: "I have underestimated the amount of expression or entertainment and creation for a lot of AI tools"
What she says is not worth her energy is lab strategy. She assumes the labs are aiming at safe artificial general intelligence with a lot of captured profit, that the efforts beneath it are ChatGPT, advertising, coding and then some expansion, and that each is worth judging on competitiveness and scope against her own companies "And I feel like people spend so much of their investing energy thinking about that." "I want to spend my energy like figuring out like okay like if we're 1% of the way in, what is the next 99% of diffusion?"
19. More AI means more work
Asked what is notably different in a year, Guo picked Jevons paradox — the pattern where making something cheaper and easier to use raises total consumption of it — applied to agents doing the mundane work. The template is software engineering, where she said the change has already happened "I have companies where they're like we are just going way faster and I expect that some analogy like that will happen"
Her example is a marketing function of one person. "The marketing department is like a person in a house" The company serves a lot of customers and needs ordinary things such as sales-enablement content. The person running marketing, she said, built an autonomous marketing department for the company instead
Asked whether she works less now that AI makes her more productive, she said the opposite. "I work more. Right. And I think this is like a core wisdom of Jensen's which is just like we're all we're all going to be more employed." Her condition on that is access: people have to be given the tooling and the education to use it
Bonus Insights
O'Shaughnessy ends every episode with the same question — the kindest thing anyone has done for you — and Guo answered collectively. What struck her about Silicon Valley is how many extraordinarily accomplished people care very little for pedigree once they have had a conversation with you, and weigh the idea and the person instead. "That is not how most ecosystems work." She started at Greylock at 23, and named Reid Hoffman, who hired her, and Joseph Ansanelli among the people who took the risk
She said young venture investors get mocked, and she hires them anyway. The criticism she quoted is that they know nothing and are a bad experience for entrepreneurs. Her answer is that her job at the time was to make other people successful, and that any task in any job can be learned by mimicry and first-principles thinking
A handful of people told her she could start the firm, which she said they would think trivial. She named Ravi Gupta, now co-chief executive of a venture called Ithaca, and Dylan Field among them, and said she would have done it either way "I will be forever grateful to I guess the people who took risk with me."
O'Shaughnessy's own closing line on what he had heard. "The world runs on faith, belief without evidence yet and still conviction in someone's ability to do something." Guo's reply was that it is faith in people, and that nobody needs a particular advantage to have an idea
Guo mentioned in passing that she hosts her own podcast, No Priors, when describing the kinds of arguments the firm has internally
O'Shaughnessy described her partner Mike as an incredibly technical person and an amazing engineer, and said the firm's younger staff bring the interesting perspectives to those debates
She said the firm hires earlier-career people deliberately, which is the same bet Greylock made on her
Guo's bottom line is that the frontier will not end up owned by one to three companies, and that what stands between the United States and cheap abundant intelligence is not the technology but the permission to build the energy and the supply chain underneath it.
Products, Companies & Tools Mentioned
Conviction (Guo's firm, founded in 2022, which invests across biology, defense, robotics and law and rates companies on a 1 to 10 scale)
Harvey (The legal AI company that made her case that law is structured language, and whose founders were projecting from a landlord-tenant question to complex M&A work)
Sunday Robotics (The robotics company whose founders she met as Stanford PhD students; under two years old and expecting semi-humanoid robots in homes, in beta, by year end)
Chai Discovery (The firm's first check, working with top-10 pharma companies on part of the R&D process — her evidence that software can make money serving pharma)
Suno (The music-generation company she was asked to invest in and declined, which she calls a mistake and evidence she underestimated demand for creation tools)
Baseten (An inference company in her portfolio whose first couple of years she said were non-obvious, and which serves a lot of open-source models)
TSMC (Named twice: as the monopoly supplier of a particular kind of glass upstream of chipmaking, and as the single production line nobody wants to depend on)
OpenAI, Anthropic and Google DeepMind (The labs she assumes are aiming at safe artificial general intelligence with a lot of captured profit, beneath which sit consumer chat, advertising and coding)
Figma, Notion and Rippling (Her examples of companies that took years to become obviously good, which is where the firm's name comes from)
Greylock (Where she started as an investor at 23, and the source of the memo-writing habit she kept)
Founders Fund (The firm she contrasts herself with: she says she has no instinct to be contrarian)
Thrive Capital (One of the entrepreneurial investment managers she admires for treating the firm itself as something you can build differently)
Books & Resources Mentioned
Sarah's Wager – Dom Cooke (The Colossus profile of Guo published in July, on her building the firm closest to the AI frontier and betting against its biggest companies; O'Shaughnessy asked whether it got her position right)
No Priors (Guo's own podcast, which she mentioned while describing the arguments the firm has about which markets are worth backing)
If this was worth your time, send it to someone closer to the industry than you are.
Get the latest market chatter as it happens: