How Cursor Grows
Cursor went from 0 to $500M ARR in 30 months. This is their story, growth playbook, and position in the market
Cursor is the fastest growing SaaS in the history of SaaS:
They reached $100M ARR in just 12 months
Then, they reached $300M ARR just 12 months later
Now, they have soared past $500M ARR (just 30 months after launch)
Despite this phenomenal growth, you’d be hard-pressed to find a well-researched breakdown of how they did it.
Enter today’s piece.
Our Roadmap
I’ve gone through everything I can find on the company for the past week to put together the web’s most comprehensive deep dive to date. We’ll cover:
Cursor’s Story
Cursor’s Growth Engine
How Cursor Builds Product
Cursor’s Position in the Market
1. Cursor’s Story
The story begins in 2020 with four close friends from MIT.
All four had been programming since they were kids.
And most had AI backgrounds.
Some worked on recommendation systems at big tech companies.
Others built robots that could learn from reward data.
One even tried to take on Google with neural network-powered search technology back in 2020-2021.
But despite their AI expertise, they weren't initially building what would become Cursor.
Chapter 1 - The Scaling Laws Revelation
In 2020, OpenAI published their scaling laws papers.
For most people, these were abstract research documents. For our 4 MIT grads, they were a roadmap to the future.
The scaling laws showed that AI capabilities would improve predictably as you scaled up data and compute. This wasn't just theoretical progress. It was predictable progress.
Around the same time, the team had a visceral moment that crystallized their belief in AI's potential. They got early access to GitHub Copilot's beta in 2021.
It was significant in 2 ways:
This was the first AI product that felt genuinely useful.
It was the most revolutionary DevTool that had been released in 10 years.
The experience of Co-pilot and the scaling laws convinced them that all of programming would eventually flow through AI models.
The question wasn't if, but when.
Chapter 2 - The False Start
Initially, the team didn't work on coding tools.
They thought that market was too crowded. Microsoft was doing it, dozens of other companies were working on it.
Instead, they picked what they thought was a "boring and sleepy" industry: mechanical engineering.
For 4 months, they built 3D autocomplete models for CAD systems, trying to predict the next changes engineers would make to 3D models.
They were essentially building a GitHub Copilot for mechanical engineering.
But there were problems from the start:
Founder-market fit was terrible. None of them were mechanical engineers.
Data was scarce. There are orders of magnitude less CAD data on the internet than code.
The technology wasn't ready. 3D machine learning was underdeveloped compared to language models.
After months of struggle, they ran out of data and momentum.
We are not mechs. There was a big blind man and the elephant problem.
Eventually, they came to their senses and returned to their original passion: programming.
Chapter 3 - The Fork Decision
When they decided to build a coding tool, they faced a critical choice: build a VS Code extension or fork VS Code entirely?
Most companies would choose the extension route. It's faster, easier, and you inherit millions of existing users. But the team made a contrarian bet.
They believed that AI would fundamentally change how people write software.
The interfaces would need to evolve dramatically.
VS Code's extensibility was too limited for the radical changes they envisioned.
From that you know it was it was clear that we were going to need to edit what the active software building looks like we were going to need to edit the UI
So they forked VS Code, a decision that seemed crazy at the time but proved prescient.
They started by building everything from scratch: their own rendering system, integrated terminal, language server integrations.
After five weeks of intense building, they were living on their own editor full-time.
Within three months of the first line of code, they launched to the public.
Chapter 4 - The Obsession with Product Quality
What happened next surprised everyone, including the founders.
They expected to build for a couple hundred users for a long time.
Instead, there was an immediate rush of interest.
But the team didn't get caught up in the hype.
They stayed obsessively focused on product quality:
We had an overriding fear of focusing on anything other than making the product better. We shunned team events, networking, or hiring to begin with," he admits. "We were all for coding on the product every day.
This became their defining characteristic.
They hired incredibly slowly, agonizing over each of the first employees.
In the end, they hired for an insane talent density.
The credentials are eye-popping:
They looked for what Stripe calls "micro pessimism and macro optimism," people who believed in the ambitious goal of automating programming but were never satisfied with the current state.
Chapter 5 - Unprecedented Growth
The obsession with product quality paid off spectacularly.
Cursor became the fastest-growing software startup in history. They reached $100M ARR in just 12 months after launching.
Then they reached $300M ARR just 12 months later—a pace of growth that has never been seen in enterprise software.
Today, they serve over half the Fortune 500, including companies like Nvidia, Adobe, and Uber.
And they are still incredibly lean. They have ~100 employees, about 90% of the team is engineers and researchers.
Chapter 6 - The Vision Ahead
For the Cursor team, this is just the beginning.
In the last month alone, they’ve shipped Cursor 1.0 and mobile agents. They just keep shipping.
Their ultimate goal isn't to build a better code editor; it's to invent an entirely new way of building software:
Our goal with Cursor is to invent a new way of programming - a very different way to build software that's kind of just distilled down into you describing the intent to the computer for what you want in the most concise way possible.
—Michael Truell
They envision a future where programming moves away from formal programming languages toward something more like natural language, but with the precision and control that professional developers need.
The idea is to shift engineering work into becoming a logic designer, where the core skill is being able to engineer the context of your intent and specifying exactly how you want everything to work.
The company’s recent raise at a near decacorn valuation (and rumors to have shut down several acquisition inquiries) shows they don’t plan to be shy on investing to capture the market.
I think that the product that we built today is just the very start of that. I think that over the next five years, we have a chance to invent a style of programming that looks very different from how you build software today.
—Michael Truell
What started as four MIT friends obsessing over scaling laws has become the fastest-growing software company in history. But they're just getting started.
2. Cursor’s Growth Engine: The 7 Layers of Cursor’s Growth
So that’s the story of Cursor, now let’s go screen-by-screen through the product and marketing to really see how this growth engine works.
You won’t find this type of deep analysis anywhere else - because no other form VP of Product at a Unicorn is writing public analyses of products at this level.
Keep reading with a 7-day free trial
Subscribe to Product Growth to keep reading this post and get 7 days of free access to the full post archives.