Windsurf is closing a huge chapter — the AI-native code editing platform is being snapped up by OpenAI for $3 billion. Today, its flagship product, the Windsurf Editor, is a fully agentic environment, not just an IDE. It’s built-in Cascade agent anticipates developer needs, fixes issues before they arise, and manages complex codebases, allowing developers to stay in flow.
According to a report from Sacra, Windsurf reached $40M in ARR in February 2025, tripling from $12M at the end of 2024. The company has expanded from 10,000 users in early 2023 to over 800,000 active developers by early 2025. The platform now processes over 100 billion tokens daily across 70+ programming languages, suggesting significant infrastructure scale and potential for continued revenue expansion.
The company recently launched Wave 8, introducing powerful tools for teams and enterprises, including AI-powered PR reviews, Google Docs integration, internal conversation sharing, direct deploys, upgraded analytics, and self-serve access controls.
Moreover, Cascade now offers deeper agentic customisation with reusable workflows, file-based rules with activation modes, simultaneous conversation support, plugin management for MCP tools, and extended multimodal integration.
However, the startup has had a fascinating journey.
From Exafunction to Windsurf
In a recent podcast, Windsurf founder Varun Mohan shared that in 2021, he and his co-founder Douglas Chen started out as a GPU virtualisation startup called Exafunction, built to make it easier to run deep-learning workloads.
But they quickly realised that with everyone moving toward Transformer-based models, their GPU infrastructure business was on track to become commoditised.
“We were managing upwards of 10,000 GPUs for a handful of companies, and we had made a couple of million in revenue, but the Transformer became very popular with models like text Da Vinci from OpenAI,” he said.
Then, in 2022, they decided to pivot the company in a different direction. Mohan revealed that the decision was made quickly, over a weekend, after a conversation between the co-founders. As early adopters of GitHub Copilot and inspired by it, they began working on Codeium, their new product, and launched it within two months.
The initial release was an AI-powered autocomplete and chat extension for Visual Studio Code (VS Code). It gained traction quickly for its speed and accuracy, but the team soon recognised the limitations imposed by existing IDE APIs.
In April this year, the company rebranded itself as Windsurf, which was initially the name of its product. It has also extended support across other IDEs like JetBrains, Eclipse, and Vim. “If we were going to work with companies like JP Morgan Chase, where half of developers use Java, we couldn’t just be a VS Code plugin,” Mohan said.
Moat is a Verb, Not a Noun
Reflecting on Windsurf’s journey, Mohan said that startups can’t rely on a single moat or advantage. “In any startup, you have to keep proving yourself,” he said. “All insights depreciate. Technology evolves, and whatever edge you had can fade quickly.”
He explained that what gives companies staying power isn’t a one-time insight but a compounding tech advantage. “If a company wins, it’s not because they had a tech insight a year ago. It’s because they’ve kept innovating and building on that insight.”
Citing NVIDIA as an example, Mohan said, “If NVIDIA doesn’t keep innovating, AMD will catch up. NVIDIA won’t be able to sustain 60–70% margins just on past wins.” Y Combinator CEO Garry Tan agreed, adding that “moat is not a noun, it’s a verb that keeps changing over time.”
Several early adopters of both Cursor and Windsurf have shared similar views. Tom Blomfield, group partner at Y Combinator, said he was initially impressed by Cursor, which “blew my mind for about 3 days,” but it started struggling as his codebase grew. He switched to Windsurf and found it handled the larger codebase efficiently, with no learning curve since both tools are VS Code forks.
While he praised both, he pointed out that they offer “zero moat or lock-in,” and users will simply move to whichever performs best at the time.
Coding Wars
Mohan agreed that the AI coding space is highly competitive right now, with tools like Cursor, Lovable, Vercel’s v0, Replit, and more in the mix.
He said that at the start of 2023, GitHub Copilot was seen as the dominant product. “Everyone would have thought GitHub Copilot was the product that everyone would use… and then came Devin… after that obviously Cursor is doing a really great job.”
Cursor reached $100 million in annual recurring revenue (ARR) within 21 months of inception. The company also recently raised $900 million in funding, led by Thrive Capital, Andreessen Horowitz (a16z), and Accel, according to The Financial Times.
On the other hand, Windsurf has raised a total of $243 million in funding. The company’s last major funding round was a $150 million Series C raised in September 2024.
Meanwhile, as of April 2025, Cursor has over 1 million users. This milestone was reached without any marketing spend, driven largely by word-of-mouth.
OpenAI also explored acquiring Anysphere’s Cursor before choosing Windsurf. Interestingly, it is also an investor in Cursor.
Amid the cut-throat competition between the two, the real winners are developers. The company’s mission remains clear, to empower developers to “dream bigger” by eliminating tedious tasks and fostering collaboration with AI. Mohan sees a future where 90% of code is AI-generated, yet engineering jobs increase due to heightened demand for innovative solutions.
