When Windsurf got picked apart in July 2025, the public record was unusually complete, and unusually unflattering to the genre. Most acquihires happen quietly, the team folded into a larger company, the product wound down, the terms never disclosed. This one was reported in detail by TechCrunch, Bloomberg, CNBC, and a former employee on Twitter, and the gap between the two accounts is the story. The founders did extraordinarily well. Many of the people who built the thing did not. That gap is the part the celebratory version of acquihire content tends to skip.
What an acquihire actually is
Strip the term down and it is the purchase of a company mainly for its people. As the Founders Network primer puts it, an acquihire is the purchase of a company “primarily to recruit its key employees,” as opposed to acquiring it for market share, intellectual property, or product strategy. The product is usually secondary and “often abandoned after the deal.” Often, the same primer notes, an acquihire is “a sign of a failing company,” and startups that get acquihired have, per CB Insights, typically raised less than $5 million.
The structure has a built-in cliff. The acquirer wants the team to stay, so the payout is back-loaded into retention packages: stock or cash that vests over time, typically with lock-up periods, as the Layerz analysis of the 2025 acquihire surge describes. The money that looks like an exit is actually a multi-year employment contract with a number attached. Leave early and you forfeit most of it. This is the mechanism that makes the experience so different from how it gets described afterward.
The Windsurf split
Windsurf was the AI coding startup formerly called Codeium, with several hundred thousand daily active users and roughly $100 million in annual recurring revenue, per Computerworld. In April 2025 it had agreed to be acquired by OpenAI for $3 billion. That deal fell apart, and in a turn TechCrunch called “shocking,” Google DeepMind hired CEO Varun Mohan, co-founder Douglas Chen, and several top researchers instead.
The terms are what make it instructive. Bloomberg reported, via TechCrunch, that Google paid roughly $2.4 billion to license Windsurf’s technology and hire its top employees. Google took no stake in the company and no control. TechCrunch named the structure directly: “the AI ecosystem’s latest reverse-acquihire, in which a company hires a startup’s top talent and licenses its technology but does not outright acquire the company.” And crucially: “Most of Windsurf’s 250 person team is not headed to Google DeepMind.” The leaders left for Google. The company, minus its leaders, was bought days later by Cognition, the maker of Devin, for undisclosed terms.
So a startup that turned down a $3 billion acquisition ended up split in two within a week: founders and a handful of researchers to one buyer, the rest of the team and the product to another. The $2.4 billion headline number attached to the people who left, not the people who stayed.
What the employees said
The most revealing account did not come from a founder. It came from an early employee, posted to Twitter and surfaced to the front page of Hacker News under the title “Windsurf employee #2: I was given a payout of only 1% what my shares where worth.” The thread drew 672 points and several hundred comments, many from engineers describing the same arithmetic from their own acquihires.
The pattern in those comments is consistent enough to be a finding. One commenter summarized the lesson bluntly: “always negotiate for higher base salaries. In the vast majority of cases, especially during acquihires, your equity will be worth little or nothing. Founders and VCs still get paid; employees rarely do.” Another described vested shares that fell from a paper value of roughly $200,000 to about $13,000 after a down round repriced everything, then a “make whole” gesture in the form of new options with a fresh four-year vesting schedule. The phrase that recurs is “money in the bank” versus “a lottery ticket.” None of this is a claim about Windsurf specifically beyond the employee #2 post; it is the texture of how acquihire participants describe the equity side once the deal closes.
The reason the gap exists is structural, not a matter of bad actors. Liquidation preferences mean investors get paid first. Founder retention packages are negotiated separately from the common stock that early employees hold. When the headline number is a licensing-and-talent fee paid to specific named people, as in the Google deal, it does not flow to the cap table at all.
The pattern repeats
Windsurf was not an anomaly. TechCrunch placed it in a sequence: Google “previously conducted a similar deal to hire back Character.AI CEO Noam Shazeer,” and Microsoft did the same “to hire Mustafa Suleyman” out of Inflection. The reverse-acquihire structure, the report notes, has “helped several Big Tech companies increase their position in the AI race without drawing regulatory scrutiny,” because no company is formally acquired.
The broader surge has a clear cause. The Layerz analysis attributes the late-2024 and 2025 spike to “tough venture conditions and an intense AI talent war”: fundraising got harder, valuations moderated, and seed and Series A startups that could not raise a follow-on round sought strategic exits, while cash-rich acquirers preferred buying proven teams to “enduring lengthy down rounds.” The Founders Network primer notes that Google alone bought 30 artificial intelligence startups between 2009 and 2020. The acquihire is not a rare last resort. It is a standing feature of how large companies hire in scarce-talent fields.
The money that looks like an exit is actually a multi-year employment contract with a number attached.
The stay-or-go decision, and whether it pays
For a founder, the deal usually comes with a question rather than a clean exit: stay for the retention period, or go. The retention package is designed to make leaving expensive, which means the choice is partly financial and partly about whether you can stand working inside the acquirer for the cliff’s duration.
Whether the bet pays is now a research question, not just a folk one. A 2025 paper in Small Business Economics, Does acqui-hiring pay off? An empirical investigation of founder retention, studies exactly this: how long acquired founders stay, and what that retention is worth to the acquirer. The full text sits behind a paywall, so we will not assert its specific findings, but the existence of a peer-reviewed study framed around founder retention confirms the central tension. The acquirer is paying for people to stay, the people are deciding whether staying is worth it, and the two calculations do not always agree. The retention cliff exists precisely because, left to their own preferences, a meaningful share of acquired founders would leave sooner than the buyer wants.
What a founder considering it should know
The honest version is not a warning against acquihires. For a founder who cannot raise and is facing a wind-down, an acquihire is often the difference between a graceful landing and a layoff. What the public record argues for is clarity about who the deal is structured to pay.
Three things hold up across the accounts. First, the headline number is rarely the number that reaches early employees. In the Google deal, it attached to named individuals and a license, not to the cap table. Second, the payout you are quoted is usually a retention package with a cliff, which means it is conditional employment, not liquidity. The commenter on Hacker News who said to “ask for the 409A valuation, liquidation preferences, and pay bands” was describing the diligence that turns a quoted number into a real one. Third, the people best positioned in an acquihire are the ones the acquirer specifically wants, because their compensation is negotiated individually rather than determined by the waterfall.
Windsurf is the cleanest case on the public record because both halves were reported: $2.4 billion to the founders and top researchers who left, an undisclosed sum to the company and team who stayed, and one early employee saying he got one percent of what he thought his shares were worth. The celebratory version of the acquihire genre tends to be written by the people in the first group. The arithmetic the second group describes is the part that does not make the press release.