Why Varun Mohan Killed a $2M Business

In 2022, Windsurf co-founder Varun Mohan was sitting on a business doing $2 million in annual revenue, backed by $28 million in funding and a team of eight engineers. Most founders would call that a win and keep going. Mohan did the opposite: he killed the business over one weekend and restarted from scratch.

That decision eventually helped turn the company, originally called Codeium and later branded around Windsurf, into one of the most closely watched AI developer tools in the market. The lesson here is simple and uncomfortable: sometimes the smartest move is to abandon the thing that is already paying the bills.

The weekend that changed everything

Mohan and his co-founder began with a very different idea. Their first company, Exafunction, focused on GPU virtualization, a technical infrastructure play aimed at making machine learning workloads cheaper and easier to run. It was a real business with real traction, not a science project. By the numbers, it had enough revenue to keep growing and enough funding to keep hiring.

Then OpenAI released GPT-3.5 and the market changed shape almost overnight. The team saw where developer attention was going, and they also saw a problem: if transformer-based models became the default, GPU virtualization would become a smaller and more commoditized business. That is a hard thing to admit when your company is already working.

Here is the part that matters. They did not spend six months debating it in public, and they did not try to preserve the old company as a side project. They took a weekend, made the call, and told the team the old direction was done.

• $28 million raised before the pivot
• 8 engineers on the team
• $2 million in annual revenue before the reset
• One weekend to decide the company’s next chapter

Why founders get trapped by good revenue

Revenue has a way of making bad strategy feel acceptable. A company doing $2 million a year can look healthy from the outside, especially if the team is small and the burn rate is under control. But if the market is moving faster than your product thesis, that revenue can turn into a trap. It buys comfort, and comfort slows decisions.

Mohan’s quote gets to the center of it: “Don’t fall in love with your idea.” That line matters because founders often confuse persistence with wisdom. Persistence is useful when the market is still validating the direction. It is a liability when the premise itself is breaking.

“Don’t fall in love with your idea.” — Varun Mohan

That mindset is rare because it asks founders to treat their own work as disposable if the evidence changes. For many teams, the hardest part is not product-market fit. It is ego management. Once a founder has raised capital, hired people, and told the story enough times, the old plan starts to feel like identity.

Mohan chose a different standard. The question was not whether the business was working today. The question was whether it could matter at the scale he wanted. That is a much stricter test.

Why Windsurf fit the moment

After the reset, the team moved into AI-powered software development. The product that emerged, Windsurf, was more than a code completion tool. It was built as an AI editor with agentic features, meaning it could understand larger codebases, help create features, and act more like a collaborator than a prediction engine.

This mattered because the market was already crowded with autocomplete-style assistants. GitHub Copilot had pushed the category into mainstream awareness, while Cursor was showing how much developers wanted AI inside the editor itself. Windsurf entered a race where the bar was no longer “can it suggest code?” The bar was “can it help me ship faster across a real codebase?”

The results were hard to ignore. Reports around the company’s growth put Windsurf’s annual recurring revenue in the range of $82 million to $100 million, a huge jump from the original $2 million business. By early 2025, the company was reportedly valued at $3 billion in a deal tied to OpenAI interest, and later the business was split across transactions involving Google and Cognition AI.

• Original business: GPU virtualization
• New business: AI coding editor and agentic workflow tool
• ARR jump: roughly $2 million to $82 million-$100 million
• Outcome: core IDE assets later acquired in a complex deal structure

The comparison that matters for founders

The easiest way to understand Mohan’s move is to compare the two paths side by side. The first path had revenue, but a shrinking strategic ceiling. The second path had more uncertainty, but it aligned with a category exploding in demand. That is the tradeoff founders face every time the market shifts.

Here is the practical comparison:

• Stay put: keep the $2 million business, preserve stability, and risk becoming irrelevant as the market standardizes
• Pivot hard: accept the short-term pain, redeploy the team, and compete in a category with much larger upside
• Split focus: try to do both and usually end up with neither
• Move fast: concentrate talent on one bet and let the old story die

That last option is the one most teams avoid because it feels wasteful. It is also the one that often creates the biggest compounding effect. A small team with strong conviction can move faster than a larger team protecting legacy revenue, especially in AI where product cycles are measured in months, not years.

There is another lesson here for anyone building developer tools. The market does not reward technical purity by itself. It rewards timing, distribution, and the ability to place your product inside a workflow people already use every day. Windsurf did that by moving from infrastructure into the editor, where the work actually happens.

What this says about AI startups now

Mohan’s pivot is a useful case study because it strips away the mythology around founders. The decision was not heroic in a cinematic sense. It was analytical, uncomfortable, and probably terrifying for the people involved. But it also matched the pace of the AI market, where waiting too long can turn a decent company into a footnote.

If you are running a startup today, the real question is not whether your product is making money. It is whether that money is buying you time to build something bigger, or just delaying the moment you admit the thesis is off. In AI, that distinction matters more every quarter.

My read: the next wave of winners will come from teams willing to cut their own best-performing product when the market gives them a better problem. If your company had to choose between protecting this quarter’s revenue and owning the next five years, which one would you actually pick?

For more founder strategy coverage, see our related piece on Netflix’s Keeper Test and what it says about hard leadership calls.