The Company That Made AI Coding Feel Inevitable

Summary

• The original Cursor thesis was interface over models — followed by a later shift for different reasons. In early 2024 the founders were “bitter lesson pilled”: “we don’t need to compete with Anthropic and OpenAI on models right now. The interface between the human and the model is the key thing,” with Michael and Aman saying code would become pseudocode. Matt summarized the product implication as “pairing the programmer’s intent down to the minimum possible spec.” Once Cursor had won users, Sarah said it had the assets, data, and know-how to build its own models — a flip that would have been impossible without starting as a frontier lab.
• The Matt-led Series A looked almost irrational against Microsoft. Copilot led “by a mile,” and Microsoft owned VS Code, the OpenAI weights, 100 million developers, and the “greatest enterprise distribution ever” — “they literally own every piece of that.” The a16z team heuristic that carried the bet: always back the leading independent in a large market, even when the incumbent seems very scary.
• The growth round required throwing out the growth model itself. Sarah Wang’s team offered to hire a sales leader at the conventional $25–50M ARR self-serve plateau; Michael “looks me dead in the eye” — “self-serve is not petering out” — forcing them to abandon assumptions like growth asymptoting to 25% in five years. The Series A was in May/early summer 2024, and Sarah co-led the B by October on “vertical liftoff”; the company was growing from 4 to 50 in four months “or whatever.”
• The founders were “paranoid but unfazed” through a rotating cast of existential competitors — Copilot, Windsurf, Cognition, and Claude Code (May 2025). Matt asked Michael about Claude Code, and Michael replied: “we are going after the biggest market in the world… you’re always going to have formidable competitors. That does not scare us” — what Sarah calls “humility mixed with bravado… the winning combo.” The genuine “oh shit” moments were model releases like Opus 4.5, and Cursor answered by cannibalizing itself “in a Reed Hastings type way”: IDE → agent platform → model platform in two years.
• Matt Bornstein dismisses the summer-2025 gross-margin discourse as an X echo chamber. “Tech transformations always precede figuring out the business” — the internet didn’t monetize for years — and investors picking apart nascent transformations have “historically been proven wrong.” The margin answer was enterprise, where labs subsidize self-serve but real margins live in third-party and enterprise, and Cursor executed “maybe the fastest build of a sales team in the history of the planet,” reaching over 50% of the Fortune 500.
• Hiring was the hidden machine: founders spending 40% of their time recruiting, AE searches run like research projects. Top 10 companies → top 10 teams → the #1 and #2 individual AEs, then “back-channel, back-channel, back-channel.” Sarah said she had never seen technical product founders spend 40% of their time recruiting. An a16z team sat in the office for eight hours at a stretch helping source.
• The endgame combination is, per Sarah, “probably the best fit I’ve ever seen” in M&A, with Matt agreeing: “Elon has the compute, they have the distribution and the data.” Sarah’s parallel is SpaceX in 2019 promising global internet with no Starlink satellites in the sky — and Cursor “beat every forecast they ever gave us… by a lot,” with the arc summed up as competing with Microsoft, then Anthropic, changing product and go-to-market, and undertaking “the most complicated M&A of all time” in two years. The acquirer is unnamed on-air.

Deep dive

1. Early 2024: interface over models, and a founder whose pitch was mostly “no”

• Matt’s scene-setting: late 2023/early 2024, the leading models were GPT-4o, Claude 3, and Llama 3, and Copilot was “the consensus thing” leading “by a mile” — no agents, no loops, no reasoning. The coding landscape ran a continuum from coding-specific base models to companies focused on agents to “a million companies doing plugins,” with Cursor “somewhere in between” — and, per Martin, “it just kind of turned out that the thing Cursor was doing was the right thing to do.”
• The signature founding conviction, quoted from Michael and Aman: “we don’t need to compete with Anthropic and OpenAI on models right now. The interface between the human and the model is the key thing.” Michael’s pseudocode argument — old CS papers state algorithms in pseudocode, and models could implement those blocks directly — “they’ve proven to be exactly right about this.” Matt summarized the product implication as “pairing the programmer’s intent down to the minimum possible spec.”
• When Michael pitched the GP group, “90% of the meeting was him saying no to things.” His reasoned case against a coding foundation model: you speak to these models in natural language, so a coding model is really a frontier language model — “just a very hard thing to do.”
• Martin’s clarity test: if you truly believe in the product, “you would never do a plugin because then you’re part of somebody else’s product,” and you defer enterprise because the product is the go-to-market. Matt’s color: asked what Michael did for fun outside work, “he didn’t understand the question” — and when Martin referenced the internet days, Michael replied “I was five.”

2. The growth bet meant discarding the growth playbook

• Martin’s setup: at the A the decision was “actually pretty obvious” — half the team used it, Karpathy was using it, “it was clearly a phenomenon… the growth was asymptotic” — but for growth investors, “you can’t write growth memos that way.” The Series A was in May/early summer 2024; Sarah co-led the B by October on “vertical liftoff.”
• Sarah’s account of conventional wisdom breaking: she offered to find a sales leader at the usual $25–50M ARR self-serve plateau, and Michael “looks me dead in the eye” and says, “self-serve is not petering out” — so the team threw out standard model assumptions like growth asymptoting to 25% in five years. “I’m very glad that we did that.”
• The competitive picture “was almost silly”: forking VS Code from Microsoft, which owned the OpenAI weights and 100 million developers and had the “greatest enterprise distribution ever.” Martin credits Matt for conviction six months before it was obvious, anchored in the house rule — “we have always bet on the independent in a large market.”
• Sarah’s data point: at a dinner Marco arranged, Aman sat next to John Schulman (“I think he was at Anthropic at the time”), whose monorepo “immediately killed” Cursor — “the next day Aman came back” and said, “we made some changes.” Signal both on iteration speed and that plugins atop VS Code “clearly [weren’t] going to work because the smartest people in the world had tried it.”

3. Paranoid but unfazed: a rotating cast of competitors, and self-cannibalization

• Martin’s admission — “there was always a looming existential threat and we almost forget about it”: Copilot, then Windsurf’s smart YC seeding, Cognition “making real waves with agents,” then the first Claude Code. Yet the founders were “entirely unfazed by the competition” — Sarah’s refinement: “paranoid but unfazed.”
• After Claude Code launched in May 2025 “with a splash,” Matt asked Michael about it. Michael replied: “we are going after the biggest market in the world… there’s been a rotating cast of characters from the very beginning… that does not scare us.” Sarah’s read: “humility mixed with bravado, and I just think that is the winning combo”; Matt calls the attitude “almost a Bezos-ish thing.”
• The real “oh shit” moments “were always around models” — Opus 4.5 landing while everyone was on vacation, and the pre-release of “Mythos [?]” Sarah says Cursor “cannibalized themselves so dramatically in a Reed Hastings type way,” moving IDE → agent platform → model platform in two years, to the point that Tab — Karpathy’s famous “tab tab tab” — “kind of doesn’t matter that much anymore.”

4. The margin discourse was noise; the enterprise flip was the strategy

• On the summer-2025 gross-margin pile-on, Matt’s dismissal: “X is driven by tail-chasing, adrenaline-seeking, and schadenfreude,” with a massive disconnect from how the business was doing. His historical frame: “tech transformations always precede figuring out the business” — the internet went unmonetized for years, and here there’s “unlimited demand, unlimited growth, and margins” — investors picking apart nascent transformations have “historically been proven wrong.”
• Sarah invokes Martin’s sequencing: “first comes the technical transformation, then comes the business model” — get all the users, then use the assets, data, and know-how to build models. “You really couldn’t do the flip of that unless you started as a frontier lab.” It was a “backdoor strategy” they “took a lot of heat for.”
• The enterprise pivot logic, per Martin: large labs subsidize the self-serve tier but command margins in third-party and enterprise, so enterprise was “existential” — beginning with literal enterprise@cursor.com — and once decided, Cursor executed “maybe the fastest build of a sales team in the history of the planet,” hitting over 50% of the Fortune 500 “faster than anyone we’ve ever seen.”
• Sarah on the AE-hiring method: research the epochs — top 10 companies for that selling motion, top 10 teams within them, then the #1 and #2 individual AEs, verified “back-channel, back-channel, back-channel” through three levels of bosses. The founders spent 40% of their time recruiting — Sarah says, “I’ve never actually seen it before” — and the a16z team sat in the office eight hours at a stretch sourcing.

5. Culture, taste, M&A — and the Elon combination

• The culture paradox Sarah flags: the founders genuinely “love writing code” — when they finally admitted to doing something for fun, “it was like, oh, that’s also coding” — yet their actual time went to recruiting, go-to-market, business models, and bizdev. Martin’s resolution: “high-powered founding teams work on exactly what is needed by the business.” The arena was possibly the most competitive market ever, because “coding is kind of the meta-category” — all expression of human thought in economic form.
• The culture showed up in details: Matt recalls no-shoes offices with slippers, hygge Scandinavian secondhand furniture, and twinkle lights; Sarah says marketing had to use “our voice” and calls the company’s follow-through on market, scope, and personnel decisions an “un-talked-about, probably greatest strength.” Martin adds that they were very good at saying no.
• On Cursor’s own acquisitions: early deals were talent plays where they were “never worried about operational complexity” and confident in setting culture; Graphite marked the shift to strategic M&A. Sarah on ex-founder integration — Tito from Koala, Adam Ward on talent — “Anthropic kind of gets that tag a lot now… but Cursor kind of did this first, and they’re actually building the company this way.”
• The final combination, which Sarah calls “probably the best fit I’ve ever seen” and Matt endorses, is a fit across technical, vision, and culture: “Elon has the compute, they have the distribution and the data,” and both believe “code is the path to changing all of compute and maybe all of humanity.” The acquirer is unnamed on-air. Sarah’s SpaceX parallel: in 2019 it was purely a launch business promising global internet with no Starlink satellites in the sky — and Cursor likewise “beat every forecast they ever gave us… by a lot,” with consensus “decrying the death of both companies so many times.”