Table of Contents
Sarah Guo of Conviction reveals the brutal economics driving VC firm growth, why "partner marketing" beats traditional brand building, and how to invest in an era where consensus ideas can actually be right for once.
Key Takeaways
- A VC firm is fundamentally just a bundle of money, people/beliefs, and competitive advantage—with money being the purest commodity
- Building brand as a new firm requires "partner marketing" with established companies to demonstrate your network rather than just claiming you have one
- The venture industry has a rational growth problem: it's easier to raise money than make money, creating oversized firms chasing returns that may not exist
- Early-stage investing is about betting on people who can lead thousands of employees, while later stages become more data-driven
- AI has created a rare moment where consensus ideas can be right, leading to multiple winners in obvious categories like code generation and customer support
- The biggest AI opportunities may be in attacking non-traditional software markets rather than existing enterprise categories
- Agency and structured reasoning remain more valuable than pure intelligence for both founders and the next generation
The Uncomfortable Truth About What VCs Actually Sell
When asked to define what a venture capital firm actually is, Sarah Guo offers one of the most honest assessments you'll hear: "Venture is a bundle of money where everybody's money is approximately green. Then you have people and beliefs that you are associated with as a founder. And then you have advantage—help you can give companies to try to make things happen faster."
The brutal reality? Money is the purest commodity. "For all tech investors love to talk about moats and differentiation and sustainable advantage, I think venture—when you think about those components, money is the purest commodity. The rest of it is pretty squishy."
This creates what Guo calls a "weird way to think about venture as a business" because the sustainable advantages are largely based on brand, relationships, and individual reputation—all things that "feel quite fragile or point in time."
Yet some firms do endure for decades. What actually compounds at places like Sequoia or Greylock? Guo identifies four key elements: ethos that persists across generations, tribal knowledge about what works, brand recognition among founders, and network effects. "I think those are the things that are harder to replicate very quickly."
Partner Marketing: The New Playbook for Emerging Managers
When Conviction launched, Guo faced a classic startup problem: nobody cared about her new firm. The solution came from a painful lesson at her previous firm, where she lost a deal to Andreessen Horowitz. The founder's feedback was brutally simple: "These things look pretty similar except they look bigger and they're better at PR and they're willing to pay more."
Guo turned this rejection into strategy. "One of the things that we wanted to do would just be demonstrate who the network is to founders in ways that are accessible to them without us having to prove it through man-to-man combat."
Enter what she calls "partner marketing"—showing up publicly with companies like Nvidia, Snowflake, and OpenAI rather than just claiming to have relationships with them. "If it was a business, it would be a funnel as part of marketing. What can we do that is efficient with people who have the distribution that we want that like us?"
This approach solves an information asymmetry problem. When founders evaluate VCs, everyone claims to have great networks and be helpful. By literally demonstrating those relationships in public forums and events, Conviction makes the intangible tangible.
The Rational Irrationality of VC Growth
Perhaps Guo's most penetrating insight concerns the economics driving the venture industry's consolidation into massive multi-stage firms. "Starting a venture firm has made me slightly more cynical about the industry," she admits.
The math is stark: "You could be a best-in-class investor delivering respectable 8 to 10 times multiple returns on a small fund and make not that much more money than somebody who's charging traditional venture fees on a fund 10 plus times your size."
This creates perverse incentives for growth over returns. Large firms can rationally subsidize early-stage investing as a sourcing funnel for later rounds, even if the early investments don't generate strong multiples. "The expectation on multiple—the cost of that capital is lower, subsidized by the growth firm to be more competitive at early stage."
The feedback loops are broken because LP money keeps flowing based on backward-looking returns from the zero interest rate era, when many funds appeared more successful than they actually were. "There's capital available to them. The feedback loop for whether or not the returns are supported are also distorted."
Guo predicts this "will not end well in terms of returns for folks," but acknowledges it's also just the natural maturation of venture as an asset class, similar to what happened in private equity.
The Art vs. Science of Investment Stages
One of Guo's clearest frameworks divides venture investing into distinct games requiring different skills. Early-stage investing remains fundamentally about people judgment, while later stages become increasingly data-driven.
"My taste is my taste and I'm an early-stage person," she explains. "I am pretty sure being great at early is going to be valuable no matter what, and I'm not globally optimizing."
At the earliest stages, you're making impossible predictions based on limited information. The core challenge: "Can this person lead a group of people—a hundred people, a thousand people? You're meeting somebody when it's just one, two people and you're trying to figure out, can this person lead a thousand people?"
This requires what she calls "force of will and the ability to take a point of view and be right." It's not just intelligence—it's the capacity to "navigate the world and shape it to what you want."
Later-stage investing follows different logic. Growth firms can wait "until the market has panned out and all the codegen companies have fought it out and one is waiting at the end," then pay a $20 billion entry price for the clear winner. Early-stage investors don't have that luxury—they have to choose before the market sorts itself out.
When Consensus Can Be Right: The AI Exception
Traditional venture wisdom says you need non-consensus ideas to generate outsized returns. AI has created a rare exception where consensus views can actually be correct, creating multiple winners in obvious categories.
"It felt to me like for a while, before AI, it felt like in order to come up with a good idea, you had to be kind of non-consensus," Guo observes. "At the moment, it feels like there's actually a lot of ideas that are both consensus and right."
This dynamic is playing out across multiple AI categories. In areas like code generation and customer support, "everybody knows it's working, they are indeed all working, and it's just not even close to the market demand yet."
The key insight is that AI is attacking fundamentally different markets than traditional software. "Some of the categories of company that we are lucky to be a part of—they attack markets that are not traditionally software markets. Harvey in law, Sierra in actual support work."
This creates pricing dynamics completely different from traditional SaaS. Instead of competing against other software at $100-200 per user per year, AI agents can price against human labor at $40,000 per year per "agent seat." "There's this very strong pricing potential that you can have" because the value proposition is replacing expensive human work rather than incrementally improving software workflows.
From Greylock's Linear Logic to AI's Nonlinear Opportunities
Guo's transition from Greylock to founding Conviction reflects a broader shift in how technology opportunities are emerging. At Greylock, she learned "linear thinking is clear thinking"—following logical patterns about how existing markets evolve with new technology platforms.
The traditional approach: "Take security or storage as an example. You look at what the transitions are that are macro technology and then you make the bets on the right types of people given that transition in an existing market."
This works brilliantly for established markets with known players and clear upgrade cycles. You understand the talent base, the spending patterns, and the adoption dynamics. "A lot of those people work at the incumbents."
But AI creates opportunities in markets that "are not traditionally software markets." The most interesting companies aren't necessarily improving existing enterprise workflows—they're creating entirely new capabilities or attacking service industries that were never software businesses.
"If you were very traditionally market focused, you might not have the time to see them in the same way," Guo explains. This requires a different approach that combines market understanding with founder-driven discovery.
The Economics of AI Abundance
The debate around AI valuations and market size gets clearer when you separate the perspectives of different market participants. When Satya Nadella says he's "looking for 7 to 10% growth" rather than chasing AGI moonshots, Guo sees rational economic decision-making.
"I think it is a statement that he does not believe that owning a particular lab or research effort that gets to a reinforcing fast takeoff is going to lead to a lot of economic value capture, or that the probability of that is not worth spending many billions of dollars."
For early-stage investors, the bar is completely different. "I have a much easier problem, which is I am quite sure that there's going to be enough economic value created to return a best-in-class venture multiple on $200 million."
The question isn't whether AI will generate enough value to justify hundred-billion-dollar infrastructure investments by hyperscalers. It's whether enough new companies will be built to generate venture-scale returns on much smaller capital bases.
Guo's optimism stems from seeing real revenue and even EBITDA at portfolio companies. "There are companies that are creating user value very rapidly and they're not spending a lot of money to do it."
The Future of Human-AI Collaboration
On the question of whether agency will remain humanity's last defensible resource as AI becomes more capable, Guo offers a nuanced view rooted in her experience evaluating founders.
"I never particularly thought intelligence was enough anyway," she notes. Even for backing founders, "besides intelligence, the thing that we most look for is kind of force of will and the ability to take a point of view and be right."
This extends to thinking about education for the next generation. While AI rapidly improves at information retrieval and even structured reasoning, Guo still believes in teaching kids "frustration management, the ability to concentrate, the ability to upskill yourself in something."
Following Andre Karpathy's reasoning, she thinks building "certain types of reasoning pathways in kids' brains" remains important, particularly around structured problem-solving. "The ability to structure a logical problem, decompose it and debug it—your ability to use even all of these tools will be better off for your ability to do structured reasoning."
But she's also "willing to be very flexible about what education looks like for them 10 years from now. If it's not Stanford, it might not be traditional education."
Building Conviction in an Uncertain World
The firm's name captures Guo's core investment philosophy: be willing to take risk on having an opinion. "Have an opinion. Maybe we're in the post-AI abundance age and venture is irrelevant, or we're on to the next big technology thing. I hope the people at the firm are on to the next big thing too."
This requires intellectual honesty about being wrong. "We published our LP letters recently. Some of those beliefs and predictions are going to be wrong and look very stupid. But for a generation of entrepreneurs right now where everybody is in a very dynamic environment and people are unwilling to make any claim about understanding or prediction, I think that can be very grounding."
The bet is that founders want investors who have developed informed opinions about where technology is heading, even if those opinions sometimes prove incorrect. In a world of rapid change, conviction matters more than perfect prediction.
Competing in the New Landscape
Despite the challenges posed by massive multi-stage firms, Guo believes smaller, specialized firms can still win at the earliest stages. "We're not trying to win everyone. We only have $200 million of capital. We need to have a couple really important companies to have a hugely winning fund in every cycle."
The key is that "people are looking for different things. The experience of working with you or working with a very specific set of people that have an opinion and have some certain understanding is quite different than working with a very large business."
It's not a homogeneous market. Different founders optimize for different things at different stages of their careers. First-time founders may care more about brand and support systems. Experienced founders like Brett Taylor "just call people they want to work with" regardless of firm affiliation.
Success comes from being excellent at what you do rather than trying to compete on every dimension. "I think great investors do this, but not me. I don't think a lot about where is the risk-reward in the mid-stage, the late stage, the early stage because I've got my game."
The venture industry's evolution toward larger, more institutionalized firms creates opportunities for nimble specialists who can offer something genuinely different. For AI-native investing in 2025, that means deep technical understanding combined with the conviction to make early bets before the market sorts itself out.
As the technology landscape continues evolving at unprecedented speed, the firms that thrive will be those that maintain clear investment philosophies while staying flexible enough to adapt to new opportunities. Conviction represents one model for how to build an enduring investment platform in an era of constant change.
