Founder Story: Varun Mohan of Windsurf
The hardest thing a founder can do is walk away from something that is working. Varun Mohan did it twice. He built a GPU infrastructure company that was generating real revenue, then shut that business down over a single realization that the ground was shifting beneath it. What he built in its place, an AI coding tool called Codeium that became Windsurf, would reach a million developers faster than almost any product in the category's history — and then be torn into pieces in a corporate drama that stunned Silicon Valley. Through all of it, one trait defined him: a willingness to abandon a good answer the moment he saw a better question.
From Sunnyvale to MIT
Varun Mohan grew up in Sunnyvale, California, the child of immigrant parents from India, in the heart of the region that would eventually shape his career. He was a standout early, competing in mathematics and computing olympiads while attending the Harker School in San Jose. The competitive-programming world is a proving ground for a particular kind of mind — fast, precise, relentless about efficiency — and it suited him.
He went east for college, earning both a bachelor's and a master's degree in electrical engineering and computer science at MIT, finishing around 2017. His studies concentrated on the unglamorous but foundational layers of computing: operating systems, distributed systems, algorithms, and performance engineering. It was there, too, that he deepened a friendship that would become a founding partnership — Douglas Chen, whom he had first met back in middle school and who became his classmate at MIT. After graduating, Mohan went to work at the autonomous-vehicle company Nuro, leading autonomy infrastructure and learning firsthand how brutally hard it is to run large machine-learning systems in production.
Building the wrong company first
In 2021, Mohan and Chen founded a company originally called Exafunction, later known as Codeium. The initial idea grew directly out of Mohan's Nuro experience: GPU virtualization infrastructure, software to help companies run machine-learning workloads more efficiently across expensive, scarce hardware. It was a sensible, technically deep business, and it worked — the company was generating meaningful revenue from real customers.
Then generative AI arrived and changed the physics of the problem. Mohan came to a conclusion that would have been easy to resist: the wave of large models about to break over the industry would eventually commoditize the very infrastructure layer his company sold. The business was profitable in the present but structurally doomed in the future. Rather than ride it down slowly, he chose to pivot hard while the company still had the strength and capital to do so. It is one thing to pivot from a failure; it is another entirely to pivot from a success.
The bet on coding
The team took the infrastructure expertise it had built and pointed it at a new target: AI-assisted software development. They launched Codeium, a free AI code-completion assistant that plugged into the editors developers already used. Offering it for free was a deliberate distribution strategy — get the tool into as many hands as possible, build a base, and worry about monetization once the product was indispensable.
Codeium grew steadily and earned a loyal following, but Mohan and Chen were not content to be one autocomplete plugin among many. They kept asking what a development environment should look like if it were designed around AI from the ground up rather than bolted onto a legacy editor. That question led to the second, larger pivot — and to the product that would make the company famous.
Windsurf and the million-developer sprint
In 2024, the team launched Windsurf, a standalone, AI-native integrated development environment. Rather than sprinkling AI features onto an existing editor, Windsurf was built so that the AI could understand a whole codebase and act across it — an "agentic" coding experience where the tool could reason about and edit projects, not just suggest the next line. The company later took Windsurf as its own name.
The response was staggering. Windsurf reached one million developers within roughly four months of launch, one of the fastest adoption curves the developer-tools world had ever seen. Mohan has been open about a lesson that surprised even him along the way: interface and experience mattered as much as raw model quality. In one telling episode, UI and UX changes alone tripled adoption overnight — proof, in his words, that for AI products the interface deserves investment on par with the model itself. The funding followed the traction, with the company raising well over $240 million and reaching a valuation above $1.25 billion.
A philosophy of discipline
For all the growth, Mohan ran the company with a striking austerity. He described Windsurf's hiring approach as keeping the organization a "dehydrated entity" — adding people only when the current team was genuinely "underwater," so that every addition was forced by real need rather than ambition or vanity. It is the same efficiency instinct that ran through his competitive-programming youth and his infrastructure work: waste is the enemy; do more with less.
His views on AI and engineering are equally sharp and, at first, seemingly contradictory. Mohan has predicted that as much as 90% of code will eventually be AI-generated, yet in the same breath argues that the number of engineers will grow, not shrink. His reasoning is that as AI improves the return on every engineering hour, companies will want more of that leverage, not less — a rising-tide view that puts him at odds with the common assumption that AI tooling means smaller teams. It is a founder's optimism grounded in economics rather than sentiment.
The seventy-two-hour breakup
Then came one of the most bizarre episodes in recent tech history. In July 2025, Windsurf — by then a coveted asset in the white-hot AI coding market — was pulled apart in a matter of days. A widely reported acquisition by OpenAI collapsed. On the evening of 11 July 2025, Google announced a roughly $2.4 billion deal to hire Windsurf's key leadership and license certain technology on a non-exclusive basis. Varun Mohan and Douglas Chen, along with a group of senior researchers, joined Google DeepMind to work on agentic coding and the Gemini effort.
Days later, on 14 July 2025, the remaining Windsurf business — its product, brand, intellectual property, and the team that stayed behind — was acquired by Cognition AI, the company behind the autonomous coding agent Devin. In the span of about seventy-two hours, a single startup had been split three ways: its founders and top talent absorbed into Google, its product folded into a rival. It was a messy, contentious ending that sparked debate about how such deals treat the employees left in the middle. But it also marked, unmistakably, that Mohan had built something the largest companies in the world were willing to move mountains to acquire.
Closing Thoughts
Varun Mohan's career is a masterclass in the discipline of letting go. Twice he walked away from a working business because he could see the future more clearly than the present rewarded him for doing. The first pivot, from GPU infrastructure to coding tools, saved the company from slow obsolescence. The second, from a plugin to an AI-native IDE, produced one of the fastest-growing developer products ever built. Neither move was forced on him by failure; both were chosen in the face of success, which is far harder.
The deeper lesson for founders is about attachment. Most people fall in love with the thing they have built and defend it past the point of reason. Mohan treated his products as means, not ends — instruments in service of a question about where software development was heading. When the answer changed, so did the instrument. That flexibility, paired with an engineer's obsession with efficiency and a clear-eyed read of where AI was taking his industry, is what carried him from a Sunnyvale childhood to the halls of Google DeepMind.
Your Story Could Be Next
Every founder in this series started with an idea and the will to build — what has changed is the leverage. Frederick gives you a team of AI agents that discover, build, and market your company, running the work across your whole business so you can focus on the decisions only you can make.
If Varun Mohan's journey resonated, there is no better time to begin your own. Start building your company with Frederick.
References
- https://www.lennysnewsletter.com/p/the-untold-story-of-windsurf-varun-mohan
- https://research.contrary.com/company/windsurf
- https://www.business-standard.com/technology/tech-news/google-hires-windsurf-ceo-varun-mohan-gemini-ai-project-openai-douglas-chen-125071200543_1.html
- https://www.computerworld.com/article/4021763/google-snatches-windsurf-execs-in-a-2-4b-deal-derailing-openais-biggest-acquisition-yet.html
- https://www.entrepreneur.com/business-news/google-inks-ai-talent-deal-with-ai-coding-startup-windsurf/494596
- https://sacra.com/c/codeium/
- https://www.crunchbase.com/person/varun-mohan-aa59