Do Reverse Acquihires Really Evade Antitrust Review?
Reverse Acquihires are all the rage in AI. Commentators say they allow an end run around the antitrust agencies. That's wrong. But they do get the acquirer something else.
Welcome back to Competition on the Merits!
There’s quite a bit of tumult in the antitrust world this week. In particular with changing personnel at the Antitrust Division. I wish Gail Slater the best of luck with whatever comes next and thank her for third stint in public service.
I also like to highlight whenever something I write here turns out to be right or eat my hat when I’m wrong. Fortunately in this case, the former is called for. The Eastern District of Texas vacated the FTC’s HSR Rule – known here as Lina Khan’s Merger Tax – on the grounds that it could not pass cost-benefit muster and was arbitrary and capricious. A year ago I predicted exactly this result. As far as predictions go, that was not a tough one. The FTC made it easy by choosing to burden all merging parties with the costs associated with a voluntary access letter level investigation – typically reserved for about 5-10% of the most problematic transactions – to all deals, i.e. even those with deals that raise no competitive issues.
This week, I want to focus on some antitrust substance in the merger world. Acquihires are all the rage. Especially when it comes to startups involving AI. We will get into definitions and explain precisely what these transactions are and are not in a moment. But for now, think about acquisitions or transactions where the primary objective is for a large tech company to acquire the key personnel at a smaller startup. It’s about the human capital more than the assets. Think Microsoft’s acquisition of Inflection. Or Google’s / Character AI. Or Nvidia’s “acquisition” of GROQ, more recently.
First, as always — please subscribe, upgrade to paid and send to friends and colleagues.
Some in Silicon Valley and commentators imagine they are a magical end-around antitrust review. Antitrust enforcers from Rohit Chopra to Jonathan Kanter to Andrew Ferguson have suggested they are suspicious and deserve investigation. Here’s Chairman Ferguson in Bloomberg recently:
Here’s the Financial Times announcing the transactions as a whole are an antitrust problem:
This skeptical view of acquihires as an end around the antitrust laws and an instrument to chill competition by gobbling up nascent competitors before they start is nothing new. Here is former Commissioner Rohit Chopra nearly 5 years ago announcing that acquihires are tantamount to unlawful collusion and a deal structure designed to avoid the Hart Scott Rodino Act. (For new readers, here’s a primer on the HSR Act):
There has been plenty of commentary about aquihires and they appear to be a growing part of the landscape with the central importance of AI. They do raise a number of really important and interesting questions. And as the agencies and commentators struggle through why we are seeing them more often, what they accomplish, and what sort of antitrust analysis is appropriate to evaluate the competitive effects of acquihires, I thought I would spend this COTM covering some of those questions:
What’s an acquihire? A Reverse acquihire?
Why are we seeing them? Is it all just an antitrust avoidance play?
How should agencies and courts think about antitrust analysis of acquihires?
Let’s start with the basics. What the heck is an acquihire?
I. Acquihires, Reverse Acquihires, and More
An acquihire is a talent-driven acquisition. It is a transaction about the human capital more than it is about any physical asset. Polsky & Coyle offer a useful definition in a 2013 law review article, describing an acquihire as “the acquisition of a firm, including all of its assets primarily for the purpose of acquiring the target firm’s employees rather than its assets of products.” That works fine to get the discussion started.
There has been a bit of an evolution in the acquihiring game. They have been around awhile. The term was coined sometime around 2005, but the concept goes back well before that. The idea is a simple one after all. It is not difficult to imagine a scenario where a large tech firm ( think Google, Meta, Microsoft, Nvidia, X, and others) wants to quickly onboard development talent. Scarcity of specific human capital is what drives these transactions. For software engineers back then. And now, mostly around AI and engineering.
The first generation of acquihires often involved full acquisitions of small startups that were frequently without significant (or any) revenue. Some of these transactions allowed smaller startups a graceful exit while helping acquirers secure engineering talent. The acquihire strategy evolved. Acquirers did not need to wait until a small start up became distressed to pick up the talent via acquisition but began to more proactively hunt for scarce skills in the marketplace. Talented teams were moved via acquisition. And those teams were the point of the acquisition rather than any intellectual property, physical assets, or customer base.
Modern acquihires are a different flavor of the same strategy. High value talent and teams are remarkably scarce in AI, robotics, security, and machine learning. But rather than an acquisition of a distressed company that was really about onboarding human capital, modern acquihires are the product of intense talent wars between large firms competing in the face to win in AI and other fields. The Reverse Acquihire is a product of this evolution. No more purchasing the entire firm and integrating two companies. The deal is about the talent. So the transaction is as well. A Reverse Acquihire takes a different form – and one that is important for our discussion later. The Reverse Acquihire usually looks a lot more like an employment contract between the acquiring firm and the talented team than it does an acquisition at all.
The Reverse Acquihire will generally consist of: (1) some compensation paid to the team – conventionally structured to create incentives for the team to stay with the acquirer for a longer period of time; (2) a non-exclusive license to the startup’s intellectual property; and (3) some kind of waiver from the target company that it will not sue the acquirer for poaching the talent. The IP portion is an important note. IP was really an afterthought in most (but not all) first generation acquihires. The acquirer might leave the startup’s IP or code behind or it might acquire it and terminate it. The IP was not the point. But in the newer acquihires, often it is the case that the talented team’s knowledge is wrapped up in code, tools, or algorithms that are the intellectual property of the target. Thus the need for some kind of license, often non-exclusive, to the IP along with the people. What is left behind at the target is an independent entity, often devoid of the talented teams that were behind its core mission. But the entity remains.
Reverse Acquihires are getting more and more common. Some examples:
Alphabet’s deal with Character AI aimed at recruiting Noam Shazeer and Daniel De Freitas
Microsoft’s recruitment of Inflection AI, which included a $650 million licensing arrangement and delivered Mustafa Suleyman to Microsoft
Amazon’s 2024 engagement with Covariant and three of it’s co-founders
Meta’s engagement with Scale AI
More recently, Nvidia’s $20 billion deal with GROQ that brings over a significant part of the target’s work force and AI inference IP but leaves the target standing as an independent entity
There are many such cases. So we have covered the phenomenon of acquihires and the rise of reverse acquihires – particularly in the AI space. The latter phenomenon is growing in frequency and, I suspect, will not go away any time soon.
Perhaps that is why it has gotten so much attention as of late from former regulators like Lina Khan and Jonathan Kanter, and at present from FTC Chairman Andrew Ferguson. So that brings us to the second set of important questions. We’ve discussed what a reverse acquihire is. Let’s turn to the why.
II. Why Reverse Acquihires? Is It All To Avoid Antitrust Review?
Answering the why acquihires question compels us to begin with the most common answer to that question: to avoid antitrust scrutiny. You see references to this explanation for the rise of the strategy everywhere. Commissioner Chopra, Chairman Ferguson, ex-Chairman Lina Khan, Jonathan Kanter – you name it – the leading explanation for the rise of reverse acquihires is that it allows the larger tech firm to avoid antitrust review and scrutiny.
I want to show my cards early here and tell you that I think this most common answer is wrong. Or at least, incomplete.
But let’s start with the conventional wisdom. The “Reverse Acquihire as Antitrust Avoidance” story goes something like this:
Chapter 1: Once upon a time, the options for startups to exit the private capital market were mergers or IPO. Sleepy antitrust enforcers never really paid attention to startup acquisitions and so that was the preferred option during the 90s and early 2000s. Fast growing startups were acquired, the FTC and DOJ slept through it, and all was well for M&A exit.
(I know there is a lot wrong with the first chapter narrative, but suspend disbelief for now)
Chapter 2: Then came Lina Khan and Jonathan Kanter. Cue the dramatic music. The Biden Administration antitrust enforcers decided to pay attention to Big Tech for the first time ever in the history of antitrust enforcement. (Keep suspending…) Khan and Kanter sued to block startup deals – raising the relative price of the M&A exit option for fast growing startups. So why not more IPOs? Increased antitrust enforcement also reduces the value of the IPO option because the post-IPO startup still wants to be acquired down the road. But the antitrust enforcers are now onto the game and won’t let them. (It is getting hard, but please keep suspending disbelief).
Chapter 3: Startups and their acquirers strategically respond to the “new” antitrust environment not with IPOs, but rather by developing and deploying the reverse acquihire. That is, startup substitution to the reverse acquihire – exodus of talent from the startup to the acquirer, coupled with a large payment, an IP license, and a waiver – is all about evading antitrust scrutiny.
This story is the one going around about the rise of reverse acquihires. I’m here to tell you it is wrong.
Well, parts of it are right. But it is simultaneously incorrect, incomplete, and misses some important antitrust details. Details worth filling out here.
Really, I think there are two interesting antitrust questions about reverse acquihires left to discuss here. The first is what they actually get you and what they do not in terms of avoidance of reduction of antitrust review. To preview: the argument that reverse acquihires buy you antitrust immunity, evade antitrust review, or anything along those lines have been dramatically, dramatically overstated. The argument has become a sort of bipartisan folk wisdom – perhaps borne out of a shared antipathy toward the Big Tech acquirers in the reverse acquihire arrangement. But it is mostly wrong for reasons we will explain. What do reverse acquirers really get? Read on. The second interesting question is a fun, substantive inquiry about how to actually analyze transactions that are entirely about adding key, scarce talent to the acquiring team rather than other kinds of assets traditionally evaluated under the antitrust framework.
III. What Do Reverse Acquihires Actually Get the Acquirer? What Don’t They?
Let’s start with what reverse acquihires do NOT get for the acquirers: antitrust immunity or any meaningful avoidance of antitrust review.
Antitrust enforcers are in no way, shape or form restricted to HSR-filed deals when conducting reviews of transactions that might violate Section 7 of the Clayton Act. The pre-merger notification process is important in that it requires companies to collect and present to the agencies some useful information that facilitates a quick first look at transactions to screen whether a more searching analysis will be needed to evaluate competition questions. There is, in the real world, always a chance that non-HSR deals will “go under the radar” and avoid agency detection. EVEN in that case, the agencies can and sometimes do sue to block consummated transactions that they missed in the first instance because no HSR was filed. Heck, if you followed the recent FTC v. Meta saga, sometimes the FTC sues to block a deal consummated a decade ago that it reviewed and cleared. The point is, one can imagine at least some meaningful correlation between deals that avoid HSR filing and deals that evade antitrust review.
So why not for reverse acquihires? The folk story that reverse acquihires buy evasion of meaningful antitrust review fails on basically every front. But let’s review a couple of the larger holes in the story.
Antitrust Agencies Did Not Discover Startup Deals in 2020. Recall, the story is that reverse acquihires are a reaction to Lina Khan and Jonathan Kanter forcing antitrust agencies to pay attention to startup deals for the first time. That increased scrutiny raised the price of normal acquisitions and thus gave birth to the cheaper alternative of the reverse acquihires precisely because it would evade both an HSR filing and meaningful antitrust review.
Not so fast, my friends. The FTC and DOJ have been paying attention to startup acquisitions for decades. “No, but Chair Khan and AAG Kanter were different,” your favorite Hipster Antitrust friend intervenes. “They did more things. They enforced the antitrust laws for the first time in 40 years.” Read the merger enforcement results and weep. They did not do more on mergers period. Not in Big Tech. Not overall. Not on a plane. Not on a train. Not on a boat. Not with a goat.
Now that does not meant the prior agencies’ record on startups is pristine. Did prior enforcement agencies miss some anticompetitive deals? Probably. Did they give a hard time to some deals that were probably procompetitive. Absolutely. Is the Biden Administration record on startups meaningfully different – other than some rhetorical flourish? Nope.
The idea that sophisticated Big Tech firms acquiring startups were scared off of “real” acquisitions to avoid the risk that the Biden DOJ and FTC – which also lost more federal court merger cases than any other Administration in the forty years prior – is far fetched. Yet, that just-so story is featured in much of the discussion of how reverse acquihires came to be, including from popular press and academia.
No HSR Filing Does Not Mean No Antitrust Review. There is a watered down version of the conventional reverse acquihire narrative though. That story goes something like: “the reverse acquihire evades HSR filing in the first place, meaning that the agencies cannot or are unlikely to sue to block a deal.”
Hogwash. This variant of the evasion story misunderstands the antitrust agencies, the HSR Act, and the Clayton Act. So let’s clear up some basics.
First, antitrust agencies review non-HSR filed transactions. It happens all the time. Does anyone really think that the FTC / DOJ did not review Microsoft / Inflection? Nvidia / GROQ? Google / Character AI? Are you kidding. I can assure you that the agencies had deal review teams in place to do a substantive antitrust review of each of these transactions.
The FTC and DOJ review non-HSR deals in all kinds of settings: rural hospital mergers, acquisitions at the local level that cause competitive problems but fall below the threshold, acquisitions below the threshold where customers complain to the agency and bring it to their attention. The idea that no HSR filing means no antitrust review is just plain wrong. So while the reverse acquihire DOES get the acquirer out of an HSR filing, there is no way that it gets out of antitrust review from the agency. In fact, I would not be surprised at all if some of the acquiring firms in the reverse acquihires of late had not made first contact with the reviewing agencies to tell them “hey, we know you are looking at this, but we are here to answer any questions you might have.” In fact, I’m certain that has happened.
The Agencies Can and Do Sue Consummated Deals. OK, so we’ve already retreated from: (1) the reverse acquihire avoids all antitrust review; to (2) it avoids an HSR filing which results in less antitrust review; to (3) a third and yet weaker variant of the evasion theory of reverse acquihires. That theory concedes that the agencies CAN and DO review non-HSR filed deals, but argues that the real import of skipping an HSR filing is that the parties can close and, as a practical matter, the agencies do not sue consummated deals. More hogwash.
FTC v. Meta, anyone? It is absolutely true that one consequence of no HSR filing is that the acquiring firm can close the deal without the HSR waiting period. That is definitional. But does that mean the FTC or DOJ cannot sue? No. They sue to block consummated deals. And there is nothing in the HSR Act or the Clayton Act that prevents the same. Nothing. If anything the hesitation to sue consummated deals as a practical matter comes from a reluctance to “unscramble the eggs” that have already merged into the proverbial omelette. But in the reverse acquihire context, the integration is limited to the talent and human capital. There really is no integration of physical assets to worry about when blocking a deal. And let’s not forget that the agencies – if they move fast when they learn of a potential reverse acquihire – can certainly run to court to get a TRO before consummation if they feel strongly about a deal’s competitive effects.
So again, no dice on the traditional evasion theory. The reverse acquihire does evade an HSR filing fee for a giant tech firm. Who cares? But it does not buy less antitrust review. And it certainly does not buy antitrust immunity. Does it force the agencies to move a little bit faster? Sure. Does it put the agencies in a position to battle a consummated deal rather than sue pre-merger? Yes. That last point is key.
But first thing’s first – let’s understand that the reverse acquihire does not evade antitrust review in any meaningful sense of the word.
So What Does The Reverse Acquihire Get the Acquirer?
We almost gave it away in the last few paragraphs. But if you’ve hung in this long you deserve a plain answer. The reverse acquihire gets the acquirer two things: (1) certainty of closing; (2) the benefits of agencies having to challenge a consummated deal. The first is easiest enough to understand from a business perspective. The overhang of uncertainty surrounding closing can be a big deal for a company. But think about the high stakes competition in the AI field. The speed with which a transaction can be closed and the talent integrated into the acquirer might be the difference between competitive life and death.
There is a benefit to speed to close that is correlated with the HSR Act waiting period here. Speed is a universal benefit here in terms of economic value created by the transaction both in absolute terms and relative to rivals. Of course evading the HSR waiting period is a benefit to this deal structure in that sense because all speed is good; and the waiting period slows things down. That is the first benefit.
But wait! That second benefit sounds ALMOST like the thing I told you wasn’t happening just above. Almost. But not quite. What the reverse acquihire does NOT get the acquirer is immunity or evasion of antitrust review because it is simply not the case that the agencies do not challenge consummated deals. Remember the whole debacle where the Khan FTC sent around “close at your own risk” letters to every merging firm? Those letters reminded companies of basic HSR and Clayton Act provisions that hold that the FTC (and DOJ) can certainly sue after closing. So the benefit is NOT that the reverse acquirer buys immunity.
So what is it? The reverse acquisition buys a change in the de facto standard of review. Courts evaluate consummated deals differently than they do proposed deals. That is just the truth. Tim Muris has a piece on how courts treat “long-consummated” deals that points some of this out. Now some of the arguments there are also about how it should be difficult for a government agency that approved a deal, subsequently consummated, to return to block it. Fair enough. But the real point is that courts in antitrust cases involving conduct that has already occurred are inclined to seek evidence of actual anticompetitive effects before condemning the conduct.
While Clayton Act Section 7 opens the doors to probabilistic arguments about future harm, incipiency and likely competitive effects, consummated deals invite practical considerations. The question is always, when evaluating competitive effects, about the counterfactual of what would have happened without the allegedly unlawful conduct. In the case of a consummated transaction courts get to observe the world both before and after the allegedly unlawful conduct. They do not observe what the world would have looked like without it. But in cases where the allegedly unlawful merger resulted in higher output or lower prices or more innovation, courts are hard pressed to condemn a transaction without compelling evidence that the unobserved counterfactual world would have been dramatically better.
And that really is the point. The business benefit is an early close. And that is most of the story. The legal arbitrage benefit is not avoiding antitrust review. And it certainly is not antitrust immunity. Rather, it is forcing the agencies to challenge consummated deals. That is something the agencies can do and have done. But what the reverse acquihire buys the acquiring firm is a practical requirement that the agencies muster up evidence of actual anticompetitive effects to have a chance at prevailing (again, FTC v. Meta, anyone?)
That difference may be subtle, but it is incredibly important. And when the FTC or DOJ hints that the reverse acquihire deal structure may violate the anti-evasion principle of the HSR Act – it should give readers pause. The question in that context is “but-for” the potential evasion of the HSR filing itself would the firms have structured the transaction this way? And I think the answer in most cases is “unequivocally yes.”
IV. Antitrust Analysis of Reverse Acquihires and Talent Transactions
We are left with one remarkably important question. How should courts analyze competitive effects of transactions that are entirely about bringing talent from Firm B to Firm A? That is probably a deeper question requiring more space than we have here to get a satisfactory framework laid out. But I do want to share some quick thoughts and return to a longer form article later on.
The set of questions that arise are interesting – especially when thinking about the remarkably scarce set of talent in AI where it is possible to imagine shifting of one talented team among firms could have significant competitive implications.
Commentators have suggested all sorts of competitive concerns. Perhaps the primary one is that reverse acquihires are “killer acquisitions” in disguise, i.e. while a reverse acquihire leaves the target firm alive as a technical matter, once the talent is stripped from it, it cannot compete and so it much the same as an acquiring firm letting the target assets whither after the merger. Those are not new issues. And have been much discussed. Most of the killer acquisition discourse involves a target firm with a successful product that has developed as a nascent competitor to the acquiring firm. That is typically not the case in reverse acquihires. The tradeoff between allowing commercialization and exit on the one hand versus the concern about suppressing innovation is well understood – if not underdocumented from an empirical perspective.
The new concerns that arise about reverse acquihires related to “talent hoarding,” i.e. the acquiring firm acts quickly to acquire scarce human capital (usually related to AI development) and preempt rivals from accessing it. Some academic work has emphasized this threat. And a handful of competition agencies have discussed it in the context of either a foreclosure-style exclusionary theory of harm (e.g., restricting rivals’ access to human capital / talent).
There are a few problems with the new aspects of these theories from an antitrust perspective that leaves me a bit dubious as to their likely impact.
The first problem is that, as a matter of law, one is hard pressed to to punish the acquiring firm for acting quickly to find talent. The literature discussing the competitive implications of talent hoarding are really about “inefficient use” of talent rather than one firm’s acquisition of talent reducing incentives to compete in the marketplace. The real concern in these models is that one firm employs a set of workers that would be more efficiently hired by a different firm – not an efficiency loss that arises out of competition itself. Here is, e.g., Benkert et al (2026):
This is fascinating as a matter of economics. But there are many inefficiencies that are outside the reach of the antitrust laws – which only address those welfare losses that do arise because of the loss of competition. This problem may be fatal for such a theory, at least under American antitrust law.
Second, the models themselves turn on sequential bidding for talent. In other words, the inefficiency arises because the acquiring firm identifies the talented team before other bidders. In particular, the better match / more efficient rival does not hire the talented team first because the incumbent is faster and recognizes the talent more quickly – or perhaps they are just lucky. Well, that is a strange definition of economic efficiency isn’t it? The incumbent is faster! It has the comparative advantage in finding scarce talent in a market where doing so really matters! But still, the talent would create more output elsewhere. OK. Fine. But the scenario where the extremely scarce talent is hard to find and a secret known only to the incumbent, but also worth billions of dollars in real world impact seems incredibly implausible as an empirical matter.
Third, when we talk about foreclosure theories and horizontal theories (for that matter) where human capital is the asset in question, things change a bit. The literature suggests that we should be MORE concerned about labor as the scarce asset because physical assets can be licensed or sold following an acquisition whereas firms do not obtain property rights over labor. Fair enough. But firms also have far less ability to keep labor in house indefinitely. Labor can move independently of the acquiring firm and the evidence is that acquihired workers leave their new employers at much higher rates than other workers. That stylized empirical fact suggests that foreclosure based concerns – i.e. a rival cannot compete because it cannot have access to the talented human capital – are less plausible in the labor context because it is harder to foreclosure access to a mobile asset. In other words, why can’t the excluded rival successfully bid for access to the human capital? Post-acquisition mobility evidence tends to suggest that reverse acquihire-specific competitive concerns are not likely as an empirical matter.
Now, the antitrust questions are not the only interesting questions here. The question of when reverse acquihires are optimal compared to alternatives is different than whether they are violations of the antitrust laws! Not all sub-optimal decisions are felonies. Thank goodness. So I do not mean to suggest there is nothing to see here in terms of interesting economic insights or differences between reverse acquihires and other deal structures. There most certainly is.
Regulators have been turned on to the idea of killer acquisitions. And certainly they are turned on to the idea that a reverse acquihire might be a killer acquisition in disguise. But beyond that idea – an idea that comes with a risk of unduly increasing the price of startup exit – what else is there in terms of antitrust substance when it comes to reverse acquihires?
Perhaps this is a bit of a wet blanket of a conclusion – but I do not think there is much. Reverse acquihires are not exciting as a substantive theory of new antitrust harm because talent hoarding largely involves potential inefficiencies that are not caused by a reduction in competition. Further, the mobility of labor post-acquihire renders the foreclosure of critical assets from rivals as even less likely than in the physical asset context.
But perhaps most importantly, the folk narrative that reverse acquihires buy the acquiring firm some form of antitrust immunity or evasion of antitrust review is simply wrong. The reverse acquihire does force the agency to litigate a consummated deal – which is well within its legal ability and experience – but requires proof of actual anticompetitive effects instead of speculation. That difference is critical when it comes to real litigation risk – and might well force the agencies to think twice before litigating. That is valuable. But it is not evasion.
See you in a few days when I’ve got a fun topic to discuss that goes back to the FTC’s ongoing battle with Southern Glazer and the Robinson-Patman Act.
Subscribe or upgrade to paid unless you want people to think that you spell antitrust with a hyphen.