May 2, 2025 · 1h 4m · sourcery
Windsurf: The Making of a Billion-Dollar AI Company | Leigh Marie Braswell, Kleiner Perkins · Sourcery with Molly O'Shea
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this episode of Sourcery, host Molly O'Shea interviews Kleiner Perkins Partner Leigh Marie Braswell to explore the rise of AI coding platform Windsurf, Kleiner Perkins' venture capital strategy, and modern frameworks for evaluating high-velocity AI startups.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Molly holds 17.5% of the talking time here. How this is scored →
Molly as informed peer 4.2 Guest teaching 4.8 Guest disagreement 0.4 Molly pushing back 1.2
speaking balance: gold is Molly, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 8:39 Deflecting acquisition rumors and market pessimism
Leigh Marie firmly shuts down speculation regarding OpenAI's rumored $3B acquisition of Windsurf, reiterating strict 'no comment' boundaries.
Hardest push from Molly ▶ 9:25 Challenging VC viability in a frozen exit environmentMolly directly asks whether venture capital as an entire asset class is doomed given severe exit headwinds and macro turbulence.
Biggest teaching moment ▶ 24:35 Exposing the margin illusion of API-reliant AI startupsLeigh Marie educates the audience on how fast-growing AI ARR can disguise deeply negative gross margins due to compounding foundation model API costs.
Molly holds their own ▶ 2:40 Synthesizing KP's 50-year generational leadership transitionMolly articulates the specific structural hazards facing legacy venture firms—such as zombie funds and cultural turnover—before framing KP's reinvention.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Molly as informed peer | Guest teaching | Guest disagreement | Molly pushing back | Why |
|---|---|---|---|---|---|---|
| Kleiner Perkins History, Fund Structure, and Team Model | 5 | 4 | 0 | 1 | Molly opens with extensive historical context on Kleiner Perkins' legacy portfolio, team transitions, and fund sizes. Leigh Marie elaborates on KP's 9-investor structure, generalist approach, and major/minor sector framework. | |
| Navigating Macro Turbulence and the Tech Exit Environment | 5 | 4 | 2 | 3 | Molly presses on exit turbulence and rumors of OpenAI acquiring Windsurf for $3B. Leigh Marie gives a firm 'no comment' on rumors and articulates KP's long-term orientation through market cycles. | |
| The Origin Story of Windsurf and Forking the IDE | 3 | 6 | 0 | 1 | Leigh Marie walks through the full backstory of Windsurf, from competitive math at MIT to ExaFunction's GPU virtualization pivot, and the pivotal decision at a JP Morgan conference to fork VS Code into Windsurf. | |
| Vibe Coding, Hyper-Growth Adoption, and New ARR Paradigms | 5 | 5 | 1 | 2 | The conversation shifts to vibe coding and the new 10x ARR milestones. Leigh Marie explains the hidden danger of negative gross margins when AI application wrappers repeatedly call expensive foundation models. | |
| Diligence Strategies, Outbound Sourcing, and AI Competitive Moats | 5 | 4 | 0 | 1 | Molly and Leigh Marie discuss diligence against fudged metrics and bubble fraud. Leigh Marie outlines her outbound-heavy sourcing philosophy and explains why development speed remains the only real moat in AI coding tools today. | |
| Sector Trends: Enterprise Automation, Scribes, and Financial AI | 4 | 5 | 0 | 0 | Leigh Marie highlights application battlegrounds such as enterprise search (Glean), medical scribes (Ambience), and untapped quantitative workflows in finance and investment banking. | |
| Career Journey at Scale AI and Spotting Talent Vortexes | 4 | 6 | 0 | 1 | Leigh Marie details her foundational years as an early engineer/PM at Scale AI under Alexandr Wang and explains how identifying 'talent vortexes' serves as her core leading indicator for venture investment. | |
| The Seedstrapped Startup Model and Enterprise AI Efficiency | 4 | 5 | 1 | 1 | Leigh Marie introduces the emerging concept of 'seedstrapped' startups that scale dramatically off seed capital without needing traditional Series A or B rounds. | |
| Poker Strategy for Deal-Flow and Future Outlook | 3 | 4 | 0 | 1 | The conversation concludes with Leigh Marie detailing her low-stakes home poker games as a networking and deal-sourcing engine, followed by an optimistic outlook on AI founders. |
Statements from this episode (21)
Kleiner Perkins deploys $800M venture fund and $1.2B growth fund
“We're currently deploying out of, this is our 21st venture fund, KP 21 it's eight hundred million dollars on the venture side, and then on the growth side, we have a 1.2 billion dollar growth fund that we call Select, Select Three.”
Kleiner Perkins operates with a core team of nine investors
“We've got this now core team of nine investors. We each have majors and minors. We're very collaborative with each other. And we're generalists sort of across early, early in growth.”
Kleiner Perkins checks range from $1M early-stage to $150M growth
“Outside of those, you know, we're tending to write checks that are at least a million dollars or up. And then it tends to be on the growth side, you know, our growth fund's a little over a billion dollars. And so we don't want to be like way too concentrated. …”
Braswell: AI coding assistants will become standard for every programmer globally
“This is gonna be a no-brainer for every programmer in the world to have one of these assistants.”
Braswell: JPMorgan inducted Codeium into its Hall of Innovation
“Every year they induct, I think one or two companies into their hall of innovation. So these are like companies that have made the sort of biggest impact on JP Morgan. And Codium was one of those companies.”
Braswell: AI 'Vibe Coding' Already Works Effectively for Non-Technical Users
“Vibe coding has become kind of this meme, but it is like so power powerful today, not just for developers, but like for everybody else too. And so I think that's been, what's, what's been most shocking to me in the last year of just AI in general is just like …”
Traditional T2D3 SaaS growth framework is obsolete in AI era
“The triple, triple, double, double, double, sass thing is like, you throw it out the window, you know, it's like, there's a new, it's like, 10 x.”
Braswell: Enterprise sales remain unautomated due to complex integration requirements
“Though I do think, you know, maybe my hot take is like, if you're an enterprise business, you there, we have not automated sales yet. We've not automated enterprise sales yet. And so I do still think if you're working with enterprises where there's just like s…”
Many fast-growing AI startups currently operate with negative gross margins
“At least today, you know, there's certainly AI businesses that yes, they're growing really fast, but the margins are quite negative. And so therefore like you have to bet on them raising more venture money until they can like figure that out, whether that's fr…”
Startup founders are inaccurately pitching non-recurring revenue as ARR
“Many, many times I have heard something that is not predictable, that is not recurring, be described as error. And that, you know, historically, you know, traditionally error is, is predictable. It's recurring. It's like signed, you know, contracts, usually th…”
Braswell: Majority of her investments come from outbound founder outreach
“I am a predominantly outbound investor. Like the vast majority of my investments have come from me reaching out to an entrepreneur and trying to basically get in front of what they're doing.”
Braswell: Speed is currently the only real moat in AI
“I think it's very unclear right now what Out of all these companies who has a moat, but I do think like speed right now is kind of the only moat that you can have.”
Braswell: Kleiner Perkins incubated and launched enterprise search startup Glean
“We invested in Glean or I should say, I mean, it was more of a, we kind of incubated Glean. So we had a big role of like getting Glean off the ground.”
Kleiner Perkins is actively seeking to fund AI customer support startups
“I think the only area that we don't have anything in right now that's like very obvious is, is is customer support. And so, you know, we have a growth fund. We're looking”
Braswell: Scale AI originally operated as an API for human labor
“Scale was the API for human labor back then.”
Braswell: Hyper-growth startups operate in chaos; calm suggests lacking PMF
“If you are not, like, feeling like you are running around all the time and everything's on fire, then, like, there's probably not a product market fit. Like, there's probably, it's not, there's not that, the only, like, hyper-growth startups, even the ones tha…”
Startups will soon jump directly from seed rounds to Series C
“Like, yes, you'll probably have to raise again, but it'll just be interesting, and I think we'll see multiple examples of, like, a company raised a seed, and then a company raised, like, a Series C.”
Kleiner Perkins uses Glean to summarize quarterly board decks
“We do this thing, portfolio review, and it's, like, every, I think every quarter. We go through all of our, or most of our investments and like glean does this thing where it will like summarize a lot of our board decks into like, okay, here's like the key met…”
Scale AI used in-office poker games as a core recruiting strategy
“At scale, we hired so many people. That was our strategy. We would bring people in the office to play poker and then, you know, like, they would get to know us and things like that.”
Braswell: AI founder momentum and tailwinds surpass 2021 tech boom
“I just like never experienced, even back in 20, 21, when there was like a very high deal volume, like I've never experienced so many smart people like wanting to become founders and like the sort of like why now and like the tailwinds being so like real. And I…”