The Windsurf Transaction and the New Reverse Acqui-hires

Live from Schenectady, New York, it's SaaS Talk with the Metrics Brothers, Growth and CAC.

And I'm Growth, better known as Ray Reich, founder and CEO of Benchmarkit.

And I'm CAC, better known as Dave Kellogg, EIR at Balderson Capital, independent consultant.

And together we are the Metrix Brothers.

And we go together like wind and waves.

Well, Dave, wind does create waves, as you know. And if you read the book The Perfect Storm and just didn't watch the movie to see George Clooney, they did do a lot of math in the book about how a big wave and the wind it can create.

Yeah, yeah, no, I love that part of the book. Quite a surprise in there to see how much wind can create waves. see all the math about how big and how steep waves get. But in fact, Ray, I was headed on a different course here, more in the windsurfing direction.

I see what you did there. Windsurf, the company that's dominated the news for the past 7 to 10 days.

Yeah, absolutely. I thought it'd be good to talk about it today because it's a very interesting tale. But more importantly, I think we should talk about the trend that it's a part of. particularly in AI, which is the new acquihires, which people are generally calling the reverse acquihires. And that's what I think we should talk about today, right?

You know, that's a great topic because while acquihires aren't really new, the new ones are certainly different from the old ones.

They are indeed, in more ways than one. So, before diving into those differences, let's hear a word from our sponsor.

The Metrics Brothers is proudly sponsored by Benchmarkit, the B2B AI and software industry's largest, broadest, and most contextualized benchmarking index. Dave Kellogg and Ray Groth Wright have put the Metrics Brothers podcast on hiatus for the remainder of 2026. All episodes we have ever recorded are still available on your favorite podcasting app. You can reach Dave @Kellogg and Ray on LinkedIn @RayReich. Thank you so much for listening to the Metrics Brothers podcast.

Okay, Ray, tell us about the deal because it's kind of a crazy tale. A lot of things happen. Just walk us through the timeline, if you would, on the Windsurf acquisition, double quotes.

Okay, I've been reading so much about this the last, you know, 7 days. I hope I can keep it at the summary level, but It all started back on May 6th when OpenAI and WindServe entered into a letter of intent, and it was for, for an acquisition, and it was basically about a $3 billion purchase. But it was subject to the existing Microsoft OpenAI strategic partnership agreement, and Microsoft kind of balked at that. And that objection on July 11th, per press, killed the deal.

So I got a question here, Ray. I'm going to break the rules and jump in. Was it Microsoft who balked on it, or was it the Windsurf people not wanting to give Microsoft access to the technology?

No, from what I read, Dave, it was Microsoft objection because they were concerned where— how OpenAI having that IP would conflict with the Copilot IP.

Oh, interesting. Okay, sorry to interrupt. So, so May 6th, OpenAI and Windsurf announced a letter of intent for an acquisition, $3 billion purchase. Then what happens?

Then July 11th, that deal fell apart, and on the same day it was announced that Google had this $2.4 billion talent plus a non-exclusive license deal. And at a high level, the 2 co-founders of Windsurf would go to Google and about 30 to 400 employees. That's not exact. And it was a $2.4 billion deal. But what's important is it left the corporate entity and the rest of the employees still at Windsurf, though the leaders were gone.

Okay. So, this is what most people might call an acquihire, although it's executed funny because in an acquihire, you typically buy the company. But here, Google, you're saying on July 11th, Google offered $2.4 billion. to hire some of the key talent and to get a license. And I've heard that referred to as a defensive license. I'm not sure if in this case it was, but in general, they just want to make sure they don't get sued. It'll give them, you know, the right to do whatever they want with the technology. So there is offensive value, but often in these deals, this defensive value— sorry to dive into that— but they offered $2.4 billion for kind of key talent and this license, be it offensive or defensive. Then what happens?

So then we, after July 11th is over, on the 12th and 13th, there was all type of online LinkedIn posts about what about the poor Windsurf employees where most of them were left in limbo? What about the entity itself? And those 2 days, a lot of noise. No one really knew quite what was going on. And then on July 14th, it was announced that Cognition, which was a rival AI coding startup to Windsurf, was going to acquire the rest of Windsurf. And importantly, because part of the noise was those people left at Windsurf, their stock didn't get accelerated, so they got nothing. But when Cognition purchased Windsurf and included 100% participation in stock by the Windsurf employees, and they accelerated their Windsurf stock to be completely vested.

Okay. So, Cognition comes along and this is an acquirer, acquires what I would call the shell company. It's not the correct use of the term shell company, but kind of the shell of a company, let's say, that was Windsurf after the key talent was poached and after there was given basically a technology license.

Got it.

So, Cognition comes along and this is rare. This doesn't happen very much where somebody comes along and buys kind of the The hustled-out shell of what remains. But it's interesting to see. Anything else? Is that the end of the story? No, one last thing.

And this is, you know, it was really from Selena Wang, who's at Altimeter Capital. The rumors are that buyout was for $200 to $300 million by cognition of the Windsurf IP and company structure, which would say about maybe $1 million to $1.5 million payout for the remaining Windsurf employees. But Dave, that has not been verified or validated by either Windsurf or Cognition. It's all just LinkedIn and press insights. But I do have a question for you. You know, we talked about how this deal went down, but let's start with what is a traditional acquihire? How is it typically structured?

Yeah, so acquihires are not a new concept in Silicon Valley. They've been around for a long time. And I'll just go to Cooley. Cooley is a famous Bay Area law firm that does corporate law. They have a nice website called Cooley Go. And I'll just tell you Cooley's description of it. An acquihire is a transaction in which a company is bought predominantly for the employees working there and not for the product service it's bringing to market. The hope is that by purchasing teams of smart people who have a history of working together, purchasers could accelerate their business growth. These transactions usually offer limited or negative returns for early investors. And usually, the acquiring companies did not continue to build and promote the acquired products and services. So basically, in English, acquihires to me were kind of a mercy killing. I don't know how else to say it. But you had a startup. It wasn't going very well. Somebody needed to come along and put this thing out of its misery. And the real value was the people. That you weren't buying a product. You were buying people. And by the way, maybe the most important metrics— we are the Metrics Brothers. is what was the metric associated with traditional acquihires? And the metric we're going to look at is basically price per head. I want to call it price. Maybe not the right word, but cost per head. And they would typically go for between $200K and $2 million an employee. So, when your company got acquihired, if you had 15 employees, you might sell it for $15 million. It might be 15 engineers. I'm not sure if those metrics are actually cost per engineer, cost per employee. But let's say— If you're in that mercy killing situation, you've fired everything but one finance person and one salesperson anyway, right? So the company is 90% engineers. So there won't be that big a difference. Go ahead.

The one thing I want to highlight that, because I have been involved in a couple of these, both as an angel investor and one as an employee. When people say it's that $250K to $2 million, or, you know, the one I was involved in was about $1.2 million per employee. That doesn't mean that that employee is getting $1.2 million. All the investors who were in the pecking order, right, with their preferences, etc., and of course, the founder who usually structured some type of payout that might have been advantageous to him or her, that's who's getting the money, not the employees.

Agree. And this will be one of the differences between the new and old acquihires. Employees don't tend to win. Rank-and-file employees tend not to win in either scenario. But in the traditional acquihire, as you noted, that's the list price, the sticker price. That's what gets written up in TechCrunch. You know, company acquired for $50 million, 15 employees, so $1 million per employee. Acquihire.

Yeah.

Ray is 100% correct that because you are acquiring a company, you're buying a company, you have a capital cap table. Basically, you have debt and you have a capitalization table. And those 2 things indicate who gets paid what. Like debt holders get paid first, always. So if that company had $5 million in debt, boom, $5 million goes to the debt provider. Now there's $10 million left. And say they had raised $20 million in venture capital. Well, there's only $10 million left. Presumably, all $10 million goes to the preferred shareholders. Now, in reality, that dog doesn't hunt 'cause the employees get nothing. So typically, the cap table does kind of get thrown out. The debt holders, I don't— I think the debt holders still go first, always.

Yeah.

I'm sure there are deals where things are screwed up enough the debt holders take a percent. But let's forget about debt holder analysis, talk about shareholders. In a traditional acquihire, the list price is a couple hundred K to $2 million. I like to do a 10x range raise. So, like $200K to $2 million per employee. But that's not just divvied up amongst the employees, as you noted. It has to kind of flow through the cap table, but then offset by, are people incented to stick around? So, in reality, as the Cooley post noted, by the way, you tend to give investors a negative return. Hey, we'll give you back $0.50 on the dollar, and that will amount to $7.5 million of the $15 million. And the other $7.5 million, we're going to give the CEO $3 million, the CRO $2 million, and split the rest amongst the 10 engineers, right? That's how it typically works. The name for the thing Ray was talking about, it's called a carve-out. Sometimes where you carve out a portion of the proceeds to go before its place in line in the cap table. And in this case, you might— it might be all carve-out. In fact, that the common stock ends up worthless, right? The preferred gets a negative return, the common gets zero. But because of this carve-out, we've got a pool of money to divvy up amongst people. And by the way, that does not include necessarily incentives offered by the new company to retain you on top of that.

Exactly right. You know, just a couple examples if the audience wants to go, you know, check out a couple recent ones. Stripe acquired Paystack, $200 million purchase price, traditional acquire, about 130 employees. So it averaged about $1.5 million per employee. And then Facebook, Meta acquired Precise. That was only a $20 million because it was such an early stage company. There were only 15 employees and that was about $1.3 million. So 2 examples that fit right in that range you talked about, Dave.

Yeah, well, they fit in the range. I'm not sure they're actually traditional acquihires. The second one to me, it looks a lot more like a traditional acquihire. The Stripe deal just looks to me like a company selling for presumably a bad price. I don't know how much money Paystack had raised. But usually to me, in a traditional acquihire, you didn't have 130 people.

Right.

You're talking about a company order of magnitude of size 5 to 20 people. So, there is a little bit subjective there. But I guess what makes it definitional on an acquihire is most of the proceeds are actually going to basically retain the staff and to buy the staff. The investors get a negative return, the debt holders get their money back. So, I'm not so sure about Stripe and Paystack. The Meta Precise one looks a lot more like what I would consider to be a traditional acquihire.

Totally agree. But today's show, we wanted to talk about this new type of acquihire. And we're going to use the Windsurf and Google deal as a way to kind of talk about that. And, you know, Dave, there's multiple names for these new acquihires. So, you think we should talk about the names before we go into the details of them?

Yeah, I think we should just for a second. So, look, the traditional acquihire, the term came from kind of acquire, but to hire all the employees, right? And you were, let's be clear, acquiring a company. And that had ramifications because if you don't acquire the company, there's still a company there and it needs to get wound down and that costs money, right? So, the traditional acquihires, you did buy a company. So, you eliminated all the wind-down costs and you kind of divvied up money amongst the survivors, for lack of a more polite term. And you gave the investors maybe a little bit back and hopefully, the debt holders got paid. That's traditional. But the— so, the acqui word belonged in there because you were acquiring But the reason you were doing it was in effect constructively, hey, let's go acquire an engineering team. Not just 10 engineers, by the way. That's why that dollar per engineering headcount is a little misleading because I do think some of the value is the fact that the teams work together for 5 years. They know each other, yada, yada. But you're buying this engineering team really more than just 10 engineers. So, the question on these new ones is what do you call them? I think the prevailing term to describe them is now reverse acquihire. And I think the argument is, oh, it's because the hire comes before the acqui, kind of. That's why it's reverse, right? And there might be other explanations if you know the derivation of it, but that's the one I saw. Somebody tried to explain why it's called reverse acqui-hire. But to me, the acqui just straight up doesn't belong in there because they're not acquisitions. There is a company left standing at the end of all this with shareholders, with a CEO, and something needs to be done with it.

Yeah. And the other, as I was doing the research for this episode, I've seen the word decoupled acquisition. I saw another one called non-acquisition acquisition. And even though you never use this as a marketer or PR person, it really is a non-acquisition acquisition. And here's why, you know, and I'll do this as a construct for one of these reverse acquihires. Number one, the primary thing you're doing is you're Buying the talent. The talent comes over. In fact, if you look at the Google purchase for $2.4 billion, they got about 30 to 40 people. Once again, it's not— hasn't explicitly been shared by either Windsurf or Google, but that's the best guess estimate from multiple people. And if it's 40 employees that went over to Google, that's about $60 million.

Wow.

Per employee, Dave, that's an average. Of course, the founders probably got a little bit more, but I was shocked at $60 million per employee because they didn't get the company. They didn't buy the company. They bought the employees.

Yeah. I mean, look, that's the— there's a couple of big differences between the new acquihires and the old acquihires. I'm just going to call them new and old acquihires. And we've already warned people that really they're not acquisitions. That's the first big difference. The old acquihires actually were acquisitions. The new acquihires aren't acquisitions. They're kind of more like hires or reverse acquihires, as a lot of people like to call them. That's the first big difference. You're not acquiring a company. The second big difference is how much you're paying per person. The old acquihires, I'm going to— you would pay on the order of $200K to $2 million. And the new acquihires, I feel like you're paying $20 million to $200 million. Right? Like, you've gone up, I think, 2 orders of magnitude. You're paying a lot. You're paying what Cristiano Ronaldo makes, right? Like, in some cases, $200 million a year. Or Dak Prescott. The script says Zach. Dak Prescott, who makes $60 million a year, right? These people are being paid like top, like the best players in the NFL or the best players in Premier League soccer. So, those are the 2 big changes. One, you're not buying a company. That has ramifications. And 2, holy cow, you're playing a lot of money. And no, Ray, I don't think it's averaging out at all. If you look at the payroll for whatever team Ronaldo's on right now, you know, he probably makes what, $200 million? The next guy might make $30 or something, right? There's a massive step down. So I think it's probably more unequally distributed than you speculated, but I'm speculating too.

Yeah, no, I totally agree. Now, in this reverse acquihire, and this isn't always in it, but I've seen it a couple times from the research. They actually got a non-exclusive license to the software, not the IP, just a non-exclusive license. And a couple of the reports said that that would be able to accelerate some of the development within Google DeepMind. But that was a little unusual to include that license, Dave.

Well, I think you may be confused here, Ray. Let me— you tell me the difference between owning something and having an unlimited non-exclusive license to it. Basically, if you give me all the source code and I can do whatever I want with it, that license effectively transfers the IP to me. Maybe I can't— I guess the only thing I can't do is exclusively sell it to someone else because the remaining shell company has it too, right? So, basically, they kept their IP as the owner, but they gave me an unlimited license to do basically whatever I want with it. Presumably, by the way, that does mean sell it to other people, though I don't know.

Yeah.

But I would just challenge you to say, wait a minute, Ray, you tell me the difference between an unlimited license to do something with a piece of software. And I can't remember, but you seem to imply they were limited in some way and they're not. They— oh, and owning the IP. I mean, to me, the only thing I think they can't do is exclusively send it to someone else because they don't exclusively own it. Right?

Right. And that's a very fine point. In one of the articles, and I wish I could source it, Right here on the top of my mind. It did say that they weren't allowed to directly sell it and be a competitor to Windsurf.

Oh, okay. Well, that would be different.

It had to be integrated into their product.

Again, you're more focused on the transaction in this episode. I'm more focused on the trend. And the trend is, I'm not going to buy— I'm not buying your company, but I want to take these 10 people and Then this is back to the offensive-defensive thing on the license. I want a license to the technology, and I could want it for one of two reasons. One, defensive, that if I own all your IP anyway, or if I have an unlimited license to all your IP anyway, you can't sue me for infringement, right? So you can't come— presumably they also include some you can't sue me for anything clause, which is probably also included. But belt and suspenders, hey, I have an unlimited license to do whatever I want with your IP. So you can't come after me. That would be the defensive license. The offensive license is, oh, I want to go sell it myself, or I want to go enhance it and sell it myself, or I want to go embed it. And I don't know in this case which they wanted. At bare minimum, you would want an unlimited license defensively. And by the way, I mean, I'm making this artificial distinction because if I have that license, I can go sell unless it precludes me selling it directly, which you could do as a separate term.

Right.

But, but just standalone, if you gimme an unlimited license to do whatever I want with your IP, I could do whatever I want with your IP, including sell it directly or embed it or license it to somebody else.

Right. And I've been involved in a couple deals where you can only directly sell the derivative rights. So you couldn't sell exactly the same exact software. There had to be derivative works and value added on top of it, and you could sell the whole thing.

Were those acquihires or were those OEM deals? What were they?

Those were OEM deals.

Yeah, because it sounds like an OEM deal to me. Okay. In an acquihire, I think you ask for more. But again, these things are all negotiated. These are big companies, billions of dollars are changing hands. So, it wouldn't surprise me if they get very nuanced on it. But the thing I know for sure in this new style of acquihire is you have to get a license to the IP. Otherwise, you could end up in court fighting about whether they stole this idea or this piece of code or this trade secret or this or that, you know, whatever it is about the software itself.

Yeah. Well, that's a really good point because there's a couple other examples of these new acquihires. You know, Google bought Character AI in August of '24 through one of these. And in fact, they actually did gain access to the underlying IP and to some of the founders. And then Amazon bought Adept. But here's the question I have. I see a lot of this going on. It's like, why, why do we see these new acquihires? And my initial perspective was it was trying to avoid the FTC and regulatory scrutiny because it wasn't, quote unquote, an acquisition. And they thought they could slide by, especially during the last FTC commissioner who was very anti-big tech acquisitions. Do you think that's directionally correct?

Yeah, I'm surprised you said that was your initial conclusion. I hope it was also your final conclusion because, yes, I think basically Life for venture capitalists got pretty hard over the past, let's say, five years plus or minus. Both doors to exit kind of got shut, right? Like the IPO window shut for reasons related to Wall Street, and then the M&A window got shut because of a more aggressive FTC. There's a great blog on this on bonalaw.com: Mergers and Acquisitions, AI, and Antitrust: The New Creative Ways for Big Tech to Enter the AI Market and Avoid HSR. Hot Scott Hart Scott Rodino. rules. And HSR is a filing that happens to happen— has to happen when you do a merger and acquisition transaction. So anyone interested in this angle should go read that blog post. But you could even hear in this post kind of the anger about how strict the FTC got. And you could certainly feel on Sand Hill Road the anger and the passion of that anger towards the FTC. So in my mind, Ray, yeah, the new acquihires are driven by a tougher M&A exit environment. I mean, somewhere in the post, it basically just says, if you're a big startup, if you're a big tech company, give up any hope of ever buying a startup because you're done, right? So these lawyers are pretty angry who wrote this thing. I presume they represent startups, or sorry, I presume they represent big tech. But anyway, yeah, this is, to me, they've created a side door Because the front door was kind of slammed shut. And by the way, I don't know how much you researched this, Ray. The FTC has not just said uncle. The FTC is still investigating these things. They're questioning whether or not they still require HSR filings anyway. And they're basically saying, hey, these are informal acquisitions, i.e., what I might call it constructively, they are acquisitions because what's really left of that remaining company. So I'm not sure this question is over by any means, but this is the current way to kind of end run the FTC to say, look, it wasn't an M&A deal at all, so we could do it tomorrow. No, no approval required, no Hart-Scott-Rodino filing, nothing. We just go.

Yeah, you know, I can't believe— I knew it. I knew it'd be hard to do this, but I did want to just focus on one last thing, and that is the key stakeholders in all this. What does this new acquihire mean to them? Hey, if you're the founders, it's a good thing, right? You're getting a hell of a lot of money. If you're a key employee, like there's approximately 30 to 40 engineers at Windsurf who went to Google, they got a hell of a lot of money. That's good. The non-key employees that stayed at Windsurf, hey, it wasn't so good for 2 or 3 days and it's still probably just okay. But then there's the investors, because you have the early stage angels and you get the VCs. What do you think, Dave? Are these new acquires good for investors?

Well, I don't want to put it that way. Let me change the question slightly on you, Ray. I think we need to look in terms of impacts because I'm not sure. I mean, in general, I don't love these things. I understand why they exist. And I think, look, I get it. I understand why they're doing them. If you want to do an acquisition, you can't do it the non-hack way. You invent a hack to get it done. Kind of the Silicon Valley, the Silicon Valley go fast and break things ethos. So I get them. I don't love that we had to create them. I find it very unsettling that the cap table is no longer the way to figure out how to unwind the company, because for years that was a simple thing. If I knew how much debt you had and I knew what your cap table looked like and somebody paid you X dollars for your company, I knew that that was the purpose in some ways of the cap table. It told me exactly how to divide that pie. And now, the cap table's kind of out the window. So, like, Dave, the angel investor, is like, gosh, I wonder what happens to the angels on these deals, especially if they're still holding a SAFE note, which is debt, right? If once they've done a valued round, then the SAFE converts into preferred stock. So, the angels become preferred stockholders, probably get treated better. But still, basically, the rules are out the window because the cap table's out the window because we're not selling the company. And the cap table says, how to divide the pie when we sell the company. But we're not selling the company. So, how do we divide the pie? And it's kind of a free-for-all, right? Between founders, key employees who will do well in my mind. And by the way, if they're not doing well, they can always say no, right? So, they're almost doing well definitionally. Investors who I think are kind of mixed on this. I'm not sure how they feel about it. In some ways, their blocking rights don't work anymore. They can't block it to my knowledge, right? And then, you've got— who did I miss, right? We got founders. Oh, you got rank-and-file employees. I guess it's founders, founders and key employees, investors, and then rank-and-file employees. I feel like I'm missing a constituency. But rank-and-file employees do not do well in these deals at all, right? So, as some people are calling it, the social contract of startups has been violated. And I don't love that as a thing. So, but yeah. So, I don't want to go as simple as good or bad on this. But I think the action item is everybody needs to understand this. If I'm joining a startup, I used to ask how much of my stock represented, and now I need to ask, and they won't be able to answer me, but I'm at least gonna have to ask about their thoughts of being acquihired and am I one of those key employees or not? So yeah, basically people and investors need to protect themselves. I'm not exactly sure what form those protections will take, but I can guarantee you in a year, SAFE notes and financing documents, I believe by a year from now, they're gonna, if not already, They're gonna start to have terms that limit this. That's my prediction.

Well, my prediction is we're gonna see more of these because time is money right now, especially in AI development and velocity of release. And if we have an FTC that's a little bit more friendly than the previous FTC, I think until something happens like some type of suit by the FTC or a class action suit, we're gonna see more of these, Dave.

Well, look, there's 2 things, Ray, to wrap this up. One, I do believe the FTC has gotten more friendly with the administration change, but these are happening anyway. So we don't need to speculate. We have proof positive that even with a friendlier FTC, these are still happening. I think basically, if the Silicon Valley view would be, hey, you forced us to create a hack to get around you, and now you can be nicer, but I'd still rather just go around you. So I think People are gonna wanna try and do these transactions. I think 2 things may limit them. One, investor protections on financing, 'cause we really don't know how the investors did on these deals. But if investors are getting hosed, they're gonna do something to prevent this. I'm not sure what they can do, not a lawyer, but investor protections could limit these, Ray. And the other thing is the FTC and the lawyers could look at these and go, this is bunk. This really is obstructively an acquisition. It needs— basically, they could slam the side door shut legally, which I think is a probability. I'm not going to say how likely, but it's certainly, I don't know, 1 in 3 chance. I mean, it's not a 1 in 100 shot. There's some pretty good arguments as to why these are constructively acquisitions, by the way, which is why to close the episode, Windsurf is actually good news for the people who like these, because for once the remaining company was worth something and got sold as opposed to just be a shell and shut down and die. Right. Because for every shell that just shuts down and dies, it reinforces the argument that these really were acquisitions.

Hey, Dave, we took a very complex and nuanced topic that Windsurf and Google kind of new acquihire, and we did a great job, I think, of talking about the old school acquihire, the new school acquihire. And I had a lot of fun researching and talking. Thank you so much.

Thanks, Ray. Take care. The Metrics Brothers, AI, SaaS, and More is a production of The Metrics Brothers Growth and CAC and a member of the Benchmarkit Podcast Network. By accessing this podcast, you acknowledge that the Metrics Brothers make no warranty, guarantee, or representation as to the accuracy or sufficiency of the information provided or the humor content of the jokes told. Ray? Information, opinions, and recommendations presented are, according to our spouses, probably wrong and provided for information purposes only. This podcast should not be considered professional or, for that matter, unprofessional advice. We disclaim any and all liability for any direct, indirect, indirect, misdirect, incidental, special, ordinary, consequential, inconsequential, Financial or other damages arising out of any use of, or God help you, reliance upon the information presented here. Ray Grothreich is based in New York City and available on X @RayReich. Dave Kellogg is based in Silicon Valley and available on X @Kelblog. Schedectady, which is French for unspellable, is not our actual production location. Thanks for listening.