"Going to Google, I'm only worth 1% of my original equity!" After the AI unicorn was divided up, the second employee angrily exposed the inside story. The founder led the way in embezzling 2.4 billion yuan and was besieged again.

Not long ago, the disintegration process of Windsurf, a star startup in the field of AI programming, attracted wide attention in the domestic and international technology circles.

Among them, the former CEO of this company led the core technology team to join Google, leaving nearly 200 employees in place, undecided about their future and full of unease. Fortunately, a few days later, another AI startup, Cognition, stepped in to take over the remaining team, temporarily ending this sudden storm. However, for the employees at the center of the storm, it was more like a "limiting choice between two" with little room for manoeuvre.

Today, Prem Qu Nair, the second - ranked employee of Windsurf, posted on the social platform X, publicly sharing his latest whereabouts and candidly recounting the impact of this sudden change on him:

I have now joined Cognition and continue to invest in the future development of software engineering.

Previously, I was the second - ranked employee at Windsurf, always working on projects related to the combination of AI and programming. The most interesting and promising aspect in this field right now is Cognition.

Actually, I could have stayed at Google DeepMind as part of the acquisition arrangement. For legal reasons, I'm not convenient to go into too many details, but I want to be honest about my choice.

At that time, I received a new offer that would expire on the same day, but the premise was that I had to give up all the equity I had obtained during my more than three - and - a - half - year tenure at Windsurf. In the end, I only received compensation equivalent to 1% of the original value of my equity.

I finally decided to go to Cognition, choosing a completely different path.

For someone who loves software engineering, Cognition feels like going back to the early days of starting a business at Windsurf - writing code like crazy and enjoying the fun immensely.

I'm really looking forward to combining the best of Devin and Windsurf to create the world's best IDE and intelligent programming assistant.

Looking back on this experience, it still makes people sigh. Windsurf's acquisition path has been full of twists and turns, with its fate changing several times. From OpenAI's high - price acquisition negotiation, to Google's "interception", and finally to Cognition taking in the remaining team, the story behind this acquisition is far more complex than what the outside world sees.

01 First Round: OpenAI Planned to Acquire Windsurf for $3 Billion, but the Deal Unexpectedly Fell Through

It's worth noting that Windsurf has only been established for four years. It was co - founded by MIT graduates Varun Mohan and Douglas Chen, initially named Exafunction, focusing on GPU virtualization infrastructure services.

In 2022, the founding team judged that generative AI was the future development trend. The company transformed into developing an AI programming assistant tool and renamed itself Codeium, positioning it as a free alternative to GitHub Copilot, which had just been released at that time.

Codeium expanded rapidly between 2022 and 2023. By the end of 2023, it had attracted more than 700,000 developers and thousands of enterprise customers. Meanwhile, they completed multiple rounds of financing, including a $65 million Series B financing, with a valuation of about $500 million at that time. Eight months later, Codeium completed a $150 million Series C financing led by General Catalyst, and the company's valuation reached $1.25 billion, making it a "unicorn" in just three years.

At the end of 2024, the company changed its name again, changing its customer - facing brand from Codeium to Windsurf. This renaming came with a grander goal: not just being a plugin or an assistant, but building a complete "AI - native" integrated development environment (IDE).

This year, Windsurf has become one of the leading independent AI programming assistants, on par with Cursor and GitHub Copilot. Although it is smaller in scale than the latter two, according to TechCrunch, as of April, Windsurf's annual recurring revenue (ARR) had exceeded $100 million.

All along, Windsurf has been on a smooth track with infinite potential. That's why its strong development momentum caught the attention of OpenAI.

In May this year, according to Bloomberg, OpenAI intended to acquire Windsurf for $3 billion to strengthen its position in the AI coding field. Meanwhile, there were also reports that Windsurf was planning a new round of financing at a valuation of $3 billion. In other words, OpenAI's acquisition proposal actually came before this round of financing.

However, this seemingly certain deal didn't progress as expected. Two months passed, and the acquisition was never officially announced. OpenAI quietly withdrew, and Google stepped in instead.

Some analysts pointed out that OpenAI's decision to withdraw might be due to considerations of regulatory risks and the constraints of its complex relationship with Microsoft. After all, Microsoft is OpenAI's major investor.

If OpenAI rashly acquired a top - tier AI IDE startup that might compete with Microsoft's technology stack or even directly "invade" its ecosystem, it would not only be embarrassing but also likely to cause turmoil in the cooperation between OpenAI and Microsoft, or even lead to a breakdown in their relationship.

So, OpenAI didn't act recklessly but strategically chose to withdraw.

02 Second Round: Google Stepped In, Paying $2.4 Billion for "Part" of Windsurf

What really triggered subsequent controversies was Google DeepMind's official "takeover" in July. However, it didn't acquire Windsurf as a whole. Instead, in a rather controversial way, it only brought in Windsurf's core team.

According to the agreement, Google DeepMind hired Varun Mohan, the co - founder and CEO of Windsurf, co - founder Douglas Chen, and some top researchers to join the team. However, it didn't take a stake in Windsurf and didn't control the company's operations. In exchange, Google obtained a non - exclusive license to some of Windsurf's technologies. That is to say, Windsurf could still license the relevant technologies to other partners.

The total price of this "package" deal of talent and technology was about $2.4 billion, and most of it went into the pockets of the shareholders.

This move was regarded as another typical "reverse acquisition" event in the AI field. That is, the tech giant doesn't directly acquire the startup but achieves low - intervention and high - integration by recruiting its core members and obtaining technology licenses. This model also appeared in Google's cooperation with Character.AI CEO Noam Shazeer and Microsoft's recruitment of Mustafa Suleyman (now the CEO of Microsoft AI).

Meanwhile, Windsurf announced that Jeff Wang, the former business leader, would serve as the interim CEO, continuing to lead most of the company's business serving enterprise customers.

It is reported that Windsurf had about 250 employees at that time. Only more than 40 researchers and engineers were poached by Google, and most of the employees were "abandoned" at the original company, continuing to operate the AI programming tool product.

Obviously, the departure of Varun Mohan, the co - founder and CEO of Windsurf, and co - founder Douglas Chen undoubtedly brought a lot of uncertainties to Windsurf's future. Referring to similar precedents in the industry, there are many cases where startups quickly lost their original advantages after losing their founding teams. For example, Scale AI suffered customer losses after cooperating with Meta, and Inflection even transformed into an enterprise service platform after its core members joined Microsoft.

At this time, Windsurf, which had been sailing smoothly, stood at a crossroads of fate.

03 Third Round: AI Startup Cognition Quickly Took Over in 48 Hours, and Windsurf Was Split in Two

Now, we already know that the AI startup Cognition quickly took over all of Windsurf's remaining employees in just 48 hours.

Cognition's acquisition terms were not made public, but according to Business Insider, this offer was obviously much lower than the $3 billion initially proposed by OpenAI. According to three people familiar with the matter, Cognition, with a market value of only about $4 billion, paid about $300 million in stock for this transaction.

Nevertheless, this "all - or - nothing" deal was far better for Windsurf employees than the outcome they feared after Google's acquisition.

Jeff Wang, the new CEO of Windsurf, posted on X afterwards, recalling: "When we announced this deal at the all - hands meeting, the employees gave a long round of applause, and I almost shed tears myself."

He wrote:

Euphemistically speaking, Windsurf had a crazy week last week. There were numerous rumors and reports outside, so I want to personally restore the true course of events.

Before I start, I want to say that Varun and Douglas are very excellent founders. This company means a lot to them, and we must admit that this whole thing must have been very difficult for them too.

Last Friday, I walked into the office, ready to attend the all - hands meeting. At that time, our about 250 employees all thought that the news of being acquired by OpenAI would be announced soon. But in fact, I already knew what was really going to happen and had told the new president, Graham, and the new CTO, Kevin, in advance. You can imagine how shocked the team was when they learned the truth.

My responsibility at that time was to explain to everyone the company's future development path. In my opinion, we had several options: we could try to continue raising funds (there were indeed venture capitalists willing to invest), we could try to sell the company (we received interest from multiple parties), we could choose to liquidate (although there were still some debt issues to deal with), or we could choose to continue operating. Although we had lost some excellent members and our morale was hit, our intellectual property, products, talents, including the first - class GTM (go - to - market) system, still remained - the core assets of the company were still there.

But the atmosphere at that time was very low. Some people were dissatisfied with the financial returns, some were sad about their colleagues leaving, and some were worried about the future. Some people even shed tears on the spot, and the atmosphere in the Q&A session understandably became a bit hostile. In the anxious mood, someone asked if we could distribute the cash on the books immediately, but we still needed the funds to pay the bills and ensure the normal operation of the products for customers.

After trying to soothe the team's emotions, that night, Graham and I started making non - stop phone calls, trying to sort out all the feasible solutions.

Just then, completely unexpectedly, we received text messages and emails from Scott and Russell, which simply said, "Let's talk?" It was around 5:30 p.m. last Friday.

I immediately called Graham and said I thought this combination made a lot of sense. Cognition has always been the team most respected by our employees. They invest heavily in engineering but obviously invest insufficiently in GTM and marketing - and this is exactly our team's strength. Meanwhile, we are currently lacking a core engineering team, and Cognition's AI engineering talents are undoubtedly among the top in the industry.

Even better is the complementary product logic: Devin is suitable for building front - end synchronous agents, while what we need is back - end asynchronous agents. After the combination of these two teams and products, it is expected to create an unparalleled end - to - end platform.

We took the contact with Cognition very seriously from the beginning and directly entered the negotiation process. Scott and his team acted quickly. While the outside world was still immersed in various memes and comments, from Friday to Monday, Scott and Russell stayed in our office almost around the clock, advancing the transaction process. On Saturday, I brought Kevin to the office. At that time, we were still having one - on - one conversations with enterprise engineers, hoping they would stay, and at the same time, we were collecting as much information as possible to evaluate Cognition and other potential partners.

During Saturday, we still received active contacts from other potential acquirers, and one of the companies was even an object we had long admired. But just then, Scott was sitting in our meeting room with a paper agreement in hand, handing me a pen. We had already decided that no other partner could be more suitable than Cognition, even though there were other excellent companies showing interest. We quickly asked lawyers to review the letter of intent (LOI) and signed the agreement on the same day - less than 24 hours after Scott's first contact.

We spent the whole day on Saturday getting to know each other's businesses. When the sun shone into the office on Sunday morning, we started to prepare for finalizing the deal. Windsurf had gone through so many twists and turns, so both Scott and I hoped that this transformation must be a real "landing".

During the whole process, we also reached a consensus - and this is also one of the important reasons why I'm sure Scott is the right partner: we must properly arrange for all of Windsurf's employees. It is their efforts and talents that have brought us to this point. They deserve rewards, and better rewards than any other option. So, this became a key part of this transaction: each employee will get a share of the dividends, all cliff clauses for option vesting will be cancelled, and the vesting process of all Windsurf equity will be accelerated.

On Sunday, a large number of lawyers from both sides flooded into the office. At Scott and Russell's request, they had to finalize the deal within 24 hours. They were on standby around the clock. We almost lived and ate in the office for the whole weekend. Some people even took a nap on the table during the negotiation breaks. The whole Sunday, our team and the lawyers worked overnight to handle the final details of the transaction. On Monday morning, we went through the agreement clause by clause again, got the board's approval, and the transaction documents were ready. One lawyer said that this was one of the fastest - completed deals they had ever seen.

At 9:30 a.m. on Monday, we officially signed the final agreement for Cognition to acquire Windsurf. Scott also informed Cognition's employees in advance, and we arranged an all - hands meeting at 10 a.m. After the traumatic announcement last Friday, Scott and Russell hoped that we could start Monday with good news, and we also hoped to handle the employees' emotions better this time. In fact, Russell even flew in overnight and actually arrived at our Austin office at the last moment.

The all - hands meeting on Monday was still held in the same meeting room as last Friday, but this time, Scott stood beside me. There were some surreal moments, but there are a few moments I'll never forget: I told everyone about the benefits we had negotiated for them: "We've decided to give you a full year of vesting... oh, and also for the 2nd, 3rd, and 4th years!" And Scott's words: "The founder should go down with the ship." The employees gave a long round of applause, and I almost couldn't hold back my tears.

Now comes the real work. We have officially become a new company. Although we still operate as two entities, there is still a lot of work to be done both internally and externally. We need to continue to work hard on team building and product refinement to achieve our common ambition. After the ups and downs of the past week, now that the story is told, we are ready to turn a new page. We will focus again and build the future of AI together.

So far, Windsurf has been split in two, heading towards different fates. However, regarding this incident, many skeptical voices have pointed the finger at the original CEO, Varun Mohan, and other founders.

Amjad Masad, the CEO of Replit, a competitor of Windsurf, publicly stated: "This breaks the social contract in Silicon Valley. It's bad news for startup employees - they will be less and less willing to join startup teams. If this is the end result, what's the point of working so hard?"

Nevertheless, some industry insiders defended Windsurf's founding team. Garry Tan, the CEO of Y Combinator, said: "Varun and his team have indeed created very excellent products