Last Updated on July 7, 2026 by Taya Ziv
There’s a version of the startup dream we’ve all repeated so many times we stopped hearing the words. You join something tiny. You take the bad salary and the scary odds because of the little number on your offer letter, the equity, the lottery ticket that says if this thing works, you work too. You’re not just an employee. You’re an owner. That promise is the whole reason anyone sane agrees to build a company with four people and no furniture.
I want to talk about the week that promise quietly broke, because almost nobody framed it as what it actually was. We covered the macro version of this story before, the part where big tech spent twenty billion dollars on AI startups and didn’t technically buy a single one of them. That piece was about the antitrust trick. This one is about the people standing in the building when the trick gets pulled.
What happened to Windsurf, in 72 hours
Windsurf was a real AI coding company. Real product, real revenue, the kind of startup employees fight to get into. Then Google came in and paid about 2.4 billion dollars to hire the CEO, Varun Mohan, the co-founder, Douglas Chen, and around forty of the staff into DeepMind. Notice the verb. Hire. Google did not buy the company. It took no equity stake. It paid a giant licensing fee for the technology and wrote enormous checks to a small set of people, and roughly half of that 2.4 billion went to compensation, a big slice of it to the two founders.
So what happened to everyone else? A lot of the newer employees, the ones whose equity hadn’t vested yet, got nothing. And here’s the detail that made my stomach turn a little when I read it. Some of the people Google did hire reportedly had their existing stock grants revoked and their vesting clocks reset to zero. You spent two years earning your ownership, and the reward for being good enough to get poached was starting the timer over.
Then the part that makes this a perfect, almost unfair little experiment. Seventy-two hours later, a company called Cognition bought what was left of Windsurf. The actual company. The codebase, the brand, the customers, the people the headlines had already written off. And Cognition did the exact opposite thing. It waived the vesting cliffs, accelerated everyone’s equity, and promised every single employee would share in the deal, no matter how long they’d been there or how much had vested.
Same company. Same week. Two completely different answers to one quiet question, the only question that actually matters to the person who took the risk: do the people who built this get paid?
This is a structure now, not an accident
If Windsurf were a one-off, it’d be a sad story and nothing more. It isn’t. This is a machine.
Between early 2024 and the start of this year, Google, Microsoft, Amazon and Meta spent more than twenty billion dollars hiring the founding teams of AI startups without buying the companies. The structure even has a name now, the reverse acqui-hire, and it was engineered on purpose to slide underneath antitrust review. You pay to license the technology, you hire the brains, and you leave the legal shell of the startup sitting there with its investors and its rank-and-file, technically still alive, actually hollowed out.
Microsoft basically wrote the playbook back in 2024 when it paid around 650 million dollars to license Inflection’s tech and hire most of its team, including Mustafa Suleyman, who walked in and became the CEO of Microsoft AI. Character.AI’s founders left for Google in a licensing deal worth something like 2.7 billion. And here’s a tell worth sitting with. After the founders left, Character.AI started paying its remaining employees a monthly cash amount equal to their unvested equity, basically to keep them from rioting. When a company has to invent a monthly guilt payment to replace the dream it can no longer deliver, the dream is already dead. They’re just holding a nice funeral.
This is also why your best people keep evaporating, which is the same force I wrote about when the AI brain drain started pulling talent straight out of billion-dollar startups. The pull at the top is so strong it bends the whole structure underneath it.
Why this lands on you, even at four people
You might be reading this thinking it’s a billionaire problem. Google, Windsurf, two and a half billion dollars, what does that have to do with a pre-seed founder and a Notion doc full of dreams.
Everything, actually. Because the reverse acqui-hire doesn’t stay at the top. The logic flows downhill, and it poisons the one asset you have that the giants don’t.
Think about what you’re actually selling when you recruit employee number three. You can’t beat Google’s salary. You can’t beat their gym, their stock, their job security. The only thing you’ve got, the entire pitch, is ownership. Come build this with me, take less cash now, and own a real piece of what we make. That handshake is your single recruiting advantage on the whole planet. It is the thing.
Now watch what the headlines are teaching everyone. They’re teaching your future hires that even when a startup “wins,” even when the founders fly off to DeepMind with life-changing money, the people who wrote the code can get a reset vesting clock and a polite goodbye. Your recruit reads the same news you do. So when you slide that offer letter across the table and point at the equity line, there’s a new voice in their head whispering, sure, and what happened to the Windsurf engineers. You didn’t break the promise. The market broke it for you, and now you’re the one standing there trying to sell it with a straight face.
My take, and the part where I might be wrong
Here’s what I actually believe. The reverse acqui-hire is not just a clever tax-and-antitrust move by big tech. It’s a temptation, and it’s coming for founders too, not just employees.
Because put yourself in Varun Mohan’s chair for one honest second. Google offers you and your co-founder a fortune to walk away clean. Your investors get made whole. The only cost is that you can’t quite bring everyone, the structure works precisely because not everyone gets the check. Are you sure you say no? I’m not sure I say no. That’s what makes it dangerous. It’s not a cartoon villain decision. It’s a very reasonable, very human decision that quietly screws the people who trusted you, and it’s being normalized into the standard happy ending right now, this year, in real time.
So the rot isn’t only that big tech invented an exit that skips the workers. It’s that the exit is so lucrative for the few at the top that it trains a generation of founders to treat their cap table as a thing you maneuver around rather than a set of promises you keep. And once a founder starts thinking that way about their own people, the startup has lost the only moral advantage it ever had over a faceless corporation.
Now let me argue against myself, because there’s a real chance I’m being a romantic here. You could fairly say I’m mourning a “social contract” that was always more myth than contract. Most startup equity was already worth zero, because most startups die, and a 409A valuation was never a paycheck. From that angle the reverse acqui-hire didn’t break the dream, it just exposed how thin the dream always was, and at least the founders and early VCs are getting something instead of the usual everyone-gets-nothing wipeout. That’s a serious argument and I won’t pretend it isn’t. Maybe a partial recovery for some beats a zero for all. But I keep landing back in the same place, because there’s a difference between a lottery ticket that loses and a lottery ticket that wins for the person next to you while yours gets shredded on a technicality. The first is just risk. The second is a betrayal with paperwork. And people can smell the difference from a mile away, which is exactly why they’ll stop buying the ticket.
What to actually do about it
If you’re a founder, three things, and none of them are “wait for Congress to fix antitrust.”
First, fix your paperwork before you ever need it. Most acceleration clauses, the single-trigger and double-trigger language everyone copies from a template, are written for a clean acquisition. An acqui-hire usually isn’t legally an acquisition. That’s the entire point of the structure. So the trigger never fires and your team’s protection turns out to be a clause that doesn’t apply to the one event most likely to happen to a hot AI startup. Sit with a lawyer and define the trigger to include a talent-and-license deal, not just a stock sale. Boring, unsexy, and it’s the difference between keeping your word and discovering you technically never gave it.
Second, stop overselling the ticket. The honest pitch in 2026 is not “this equity will make you rich.” It’s “here is the real range of outcomes, including the ugly one where I get bought for my brain and you get a reset clock, and here is specifically what I’ve put in writing to stop that from landing on you.” A recruit who hears that trusts you more, not less. The discipline is the same one behind why your Series A hiring strategy is quietly burning cash, which is that honesty about money up front is cheaper than resentment later.
Third, decide who you are now, while it’s free. Right now, before any offer exists, before there’s a number with that many zeros sitting in front of you clouding your judgment, decide whether you’re a Google or a Cognition. Write it down. Tell your team. Pre-commit, out loud, that if the day comes when someone wants to buy your people’s talent and leave their ownership behind, you take the team with you or you don’t take the deal. It is so much easier to be a good person before the briefcase is open.
The handshake was the company
I keep coming back to those 72 hours, because they’re the cleanest moral test I’ve seen in years. Google had the chance to take care of everyone and chose not to, because the structure paid more if it didn’t. Cognition had the same company in front of it three days later and chose to bring everyone along. Nobody forced either hand. They just had different ideas about what they owed the people in the room.
The startup world runs on a promise that’s never written quite plainly enough: take the risk with me and we’ll share whatever we catch. The reverse acqui-hire is the first exit structure that’s profitable specifically because it breaks that promise. If you’re building something right now, the headlines are going to keep tempting you to think of your cap table as a problem to route around. Don’t. That promise is not the overhead of your company. For the first few years, it is your company. Break it the way Google did and you might get one fat check. Keep it the way Cognition did and you get the only thing that ever actually compounds, which is people who’d build with you again.
