Introduction

In March 2024, one of Silicon Valley’s promising early artificial intelligence entrants effectively unraveled without ever being formally acquired.¹ Microsoft announced that it had hired Mustafa Suleyman, co-founder of Inflection AI, along with many of the company’s key employees, while simultaneously entering into a $650 million licensing arrangement for Inflection’s technology.² The deal drew attention because Inflection had not been viewed as a struggling startup in need of an acquisition exit, but as a formidable early entrant in the AI race.³ Less than a year earlier, the company had raised approximately $1.3 billion at a $4 billion valuation.⁴ Microsoft never acquired Inflection.⁵ The company remained legally independent, even as much of its core talent and technological capacity migrated to one of the world’s most powerful technology firms.

Similar arrangements have since proliferated across the AI industry.⁶ Amazon entered into a comparable agreement with Adept AI, licensing the startup’s technology while hiring its co-founders and part of its team.⁷ NVIDIA entered into a high-profile licensing-and-hiring arrangement with AI inference-hardware company Groq, which Forbes described as “a merger without the paperwork.”⁸ Meta similarly hired Scale AI founder Alexandr Wang to lead its AI operations while investing approximately $14.3 billion for a 49% stake in the startup.⁹ These transactions are among a growing category of arrangements sometimes called “reverse acquihires,” in which a firm obtains key startup personnel and technological capabilities through some combination of hiring, licensing, investment, or partnership rather than acquiring the company outright.

Reverse acquihires raise an emerging question for antitrust law because their competitive significance may not be captured by the formal structure of the transaction. A startup may remain legally independent while losing personnel, technology, or other capabilities that could have enabled it to develop into an independent competitive constraint. Since these arrangements may not involve the type of acquisition that triggers Hart-Scott-Rodino (HSR) Act premerger reporting requirements,¹⁰ they may avoid the ordinary procedural mechanism that brings large acquisitions to the attention of antitrust enforcers.

This article explores reverse acquihires through the lens of existing Section 7 doctrine. Section 7 of the Clayton Act has long required a forward-looking inquiry into whether an acquisition “may be substantially to lessen competition,”¹¹ reflecting Congress’s concern with preventing threats to competition before they fully materialize. That predictive orientation may be particularly relevant to reverse acquihires, where a startup’s competitive significance may lie less in its present market share than in its potential to develop into a future competitive constraint. Although reverse acquihires present a novel transaction structure, the questions they raise about incipient and future competition are familiar ones in antitrust law.

Reverse Acquihires and Future Competition

The potential competitive significance of a reverse acquihire could be difficult to see through conventional measures of present competition. A startup might have little revenue, no meaningful present-day market share, or no fully commercialized product.¹² Its significance may instead lie in its future potential: the possibility that its personnel, technology, or research trajectory could develop into a meaningful competitive constraint.

Reverse acquihires can affect that trajectory without eliminating the startup as a legal entity. When an incumbent hires a startup’s founders or core technical team while simultaneously obtaining access to its technology, the startup may remain formally independent but lose some capabilities that supported its original path to market. In innovation-driven industries such as artificial intelligence, those capabilities may include engineering talent, model architectures, research pipelines, computing resources, and technical expertise. Therefore, the competitive question may extend beyond what the startup currently sells to what competitive capacity remains after the transaction.

This distinction is particularly relevant to nascent firms. Startups, almost by definition, may have small or nonexistent market shares. Their competitive importance can instead depend on whether they possess a plausible path toward becoming an independent rival. A reverse acquihire may alter that path by moving key personnel or technology into an incumbent’s organization while leaving behind a smaller company that continues to exist independently. At the same time, such arrangements can have procompetitive explanations and effects: licensing can help to commercialize technology; partnerships may provide startups with resources they could not obtain independently; and the movement of technical personnel can facilitate innovation. Hence, the existence of a reverse acquihire does not itself establish competitive harm.

The challenge is one of prediction. Evaluating the competitive significance of a startup requires considering a counterfactual that cannot be observed with certainty: how the firm and the market might have developed absent the transaction. A promising startup may fail, pivot, or ultimately complement an incumbent rather than challenge it. However, it may also develop into a significant source of future competition. Reverse acquihires thus raise a familiar question in an unfamiliar transactional setting: how antitrust law evaluates a potential loss of competition before that competition has fully emerged.

That question leads naturally to Section 7 of the Clayton Act. Unlike legal standards that depend upon demonstrated anticompetitive effects, Section 7 expressly asks whether the effect of an acquisition “may be substantially to lessen competition.”¹³ Accordingly, the U.S. Supreme Court has highlighted that Section 7 analysis is not usually “susceptible of a ready and precise answer.”¹⁴ Merger analysis requires not merely an assessment of a transaction’s immediate effects, but a “prediction of its impact upon competitive conditions in the future.”⁴ Section 7’s predictive character thus provides a natural doctrinal backdrop for considering transactions whose competitive significance may likewise lie in the future.

Section 7’s Predictive Framework

Section 7 of the Clayton Act is inherently forward-looking. The statute prohibits acquisitions whose effect “may be substantially to lessen competition, or to tend to create a monopoly.”¹³ The Supreme Court has repeatedly understood that language to make merger analysis an inquiry into probabilities rather than certainties. In Brown Shoe Co. v. United States, the Court explained that Congress amended Section 7 to address anticompetitive tendencies “in their incipiency,” giving courts and enforcement agencies the ability to respond to concentration “at its outset and before it gathered momentum.”¹⁵ Thus, Section 7 does not require enforcers to wait until a transaction’s competitive effects have fully materialized.

That preventive purpose necessarily requires some prediction about the future. In United States v. Philadelphia National Bank, the Supreme Court recognized that Section 7 analysis is not ordinarily “susceptible of a ready and precise answer.”¹⁶ Instead, evaluating a transaction requires “not merely an appraisal of the immediate impact of the merger upon competition, but a prediction of its impact upon competitive conditions in the future.”⁴ Brown Shoe similarly emphasized that Congress was concerned with “probabilities, not certainties” and therefore prohibited transactions with a “probable anticompetitive effect” rather than requiring proof of a competitive harm with certainty.¹⁷

Later merger cases made that predictive inquiry more empirically grounded without abandoning Section 7’s concern with future competition. For example, United States v. General Dynamics Corp. illustrates that present market shares may not accurately reflect a firm’s future competitive significance. The Supreme Court allowed defendants to rebut structural evidence by showing that the acquired firm lacked meaningful future competitive strength.¹⁸ There, historical production statistics overstated the acquired coal producer’s competitive strength because its depleted and committed coal reserves left it, “in terms of probable future ability to compete,” a substantially less significant competitor than its historical production suggested.⁴ The Court looked beyond a snapshot of existing market structure to evidence bearing on the firm’s actual capacity to compete going forward.

The D.C. Circuit’s decision in United States v. Baker Hughes Inc. likewise treated concentration statistics as the beginning, rather than the end, of the competitive analysis.¹⁹ Under the court’s burden-shifting framework, defendants may rebut a prima facie case based on concentration by producing evidence that market-share statistics inaccurately predict the transaction’s probable competitive effects.²⁰ The strength of the evidence required to make a successful rebuttal depends in part on the strength of the government’s initial showing.⁴ Accordingly, modern merger analysis remains predictive, but the prediction is disciplined by evidence about the particular firms, market conditions, entry, and other facts bearing on how competition is likely to develop.

These principles are relevant to reverse acquihires because the same temporal question can arise in the opposite direction. General Dynamics demonstrates that current market data can overstate a firm’s future competitive significance when firm-specific evidence shows that its ability to compete is declining. A reverse acquihire may present the converse problem: current market data can understate the significance of a startup whose existing sales or market share are small but whose personnel, technology, or research trajectory could support greater competition in the future. In either setting, Section 7’s predictive inquiry directs attention beyond present market position to evidence of probable future competitive capacity.

More recently in 2023, FTC v. Illumina, Inc. demonstrates that this forward-looking inquiry can extend to competition occurring before products reach commercialization.²¹ The Fifth Circuit upheld the FTC’s definition of a market for the “research, development, and commercialization” of multi-cancer early detection (MCED) tests, rather than limiting the market to commercialized products.²² Although only one firm had begun commercializing an MCED test, numerous firms in this “nascent field” were developing products with the goal of eventual commercialization and were already competing to bring those products to market.²³ The court rejected the argument that the market should be defined based only on products that currently existed, reasoning that such an approach could effectively “prevent research-and-development markets from ever being recognized for antitrust purposes.”²⁴ Illumina thus demonstrates that Section 7 can recognize competition at the R&D and pre-commercial stage, before products have matured into commercial rivalry, when the record supports a likely effect on future competition.

Taken together, these cases provide examples for evaluating reverse acquihires without requiring certainty about a startup’s counterfactual future. Reverse acquihires present a novel factual setting, but raise a familiar temporal problem: whether an emerging firm would have developed into a meaningful independent competitor and whether the transaction materially changes that trajectory. Section 7 does not eliminate the uncertainty inherent in that inquiry, but its longstanding focus on incipient and probable competitive effects provides a doctrinal framework for evaluating competitive threats before their consequences become fully observable.

Reverse Acquihires Through a Section 7 Lens

The structure of recent reverse acquihires illustrates why Section 7’s focus on future competitive conditions may be relevant even when a transaction does not resemble a conventional merger. Consider NVIDIA’s arrangement with Groq, an AI inference-hardware company founded by Jonathan Ross, one of the engineers behind Google’s tensor processing unit. Rather than acquiring Groq outright, NVIDIA reportedly entered into a licensing arrangement for Groq’s technology and hired Ross and other members of the company’s leadership and engineering team, while Groq continued to operate as an independent company.²⁵ News outlets reported that human capital was “[p]erhaps the most valuable asset in this deal”⁴ and even characterized the transaction as “a merger without the paperwork: take the team, secure the tech, and get the strategic benefit without inheriting every loose end in the corporate attic, or running into antitrust issues.”⁸

The transaction illustrates the distinction between formal corporate ownership and the competitive capabilities that can move between firms. For a technology startup, those capabilities may include not only existing products or physical assets, but also specialized personnel, technical know-how, intellectual property, and ongoing research. A transaction that transfers some combination of those capabilities can therefore change the firms’ respective competitive trajectories even while both corporate entities continue to exist. Public reporting alone cannot establish the competitive effect of such a transaction, but the structure raises the type of forward-looking question contemplated by Section 7: what role were the transferred capabilities likely to play in future competition?

That inquiry is necessarily fact-specific. A startup’s valuation, financing history, technical capabilities, product roadmap, and recognition by customers or industry participants may provide information about its prospects, as may ordinary-course evidence concerning how the firms themselves understood the startup’s competitive significance. The transaction’s effect on the startup can also matter. If the company retains the personnel, technology, financing, and strategic independence necessary to continue pursuing its pre-transaction trajectory, its continued development may look different from that of a company that loses the capabilities around which its future products were being built. None of these facts alone answers the competitive question. Together, however, they may provide evidence relevant to the prediction Section 7 requires.

The difficulty of making that prediction is not unique to reverse acquihires. Meta’s acquisitions of Instagram and WhatsApp illustrate the uncertainty surrounding emerging competitors. Years after those acquisitions closed, the FTC and a coalition of 46 states, the District of Columbia, and Guam brought monopolization actions alleging that Facebook had acquired emerging competitive threats as part of a strategy to maintain its monopoly in personal social networking. Although those transactions were conventional acquisitions rather than reverse acquihires, they presented a similar predictive question: how to assess an emerging firm’s future competitive significance before it has fully materialized.

The complaints relied in part on contemporaneous internal communications reflecting Facebook’s assessment of those threats. Mark Zuckerberg allegedly observed that features from apps like Instagram offered “ways for those apps to replace [Facebook]” and described emerging mobile-app companies as “nascent” businesses whose established networks and brands, “if they grow to a large scale,” could become “very disruptive to us.”²⁶ The complaint further alleged that in emails Zuckerberg stated that, by acquiring Instagram and other emerging rivals, “what we’re really buying is time”: buying Instagram, Path, and Foursquare would give Facebook “a year or more to integrate their dynamics before anyone can get close to their scale again.”²⁷ The FTC complaint also contained alleged quotes from Zuckerberg’s message to a colleague after the Instagram acquisition was announced: “I remember your internal post about how Instagram was our threat and not Google+. You were basically right. One thing about startups though is you can often acquire them.”²⁸

The subsequent litigation also illustrates the difficulties of evaluating such claims after markets have evolved. In the FTC’s case, the district court ultimately rejected the agency’s proposed personal-social-networking market after trial, finding that services offered by Meta, TikTok, and YouTube had converged over time.²⁹ However, that later convergence does not answer Section 7’s ex-ante question: what competitive developments were reasonably probable based on conditions existing at the time of the transaction? Nor does it establish what would have happened had Instagram or WhatsApp remained independent. The litigation thus highlights the counterfactual problem inherent in evaluating emerging competition ex post: both the firm and the market may look substantially different years after a transaction, potentially obscuring the competitive possibilities that existed when it occurred.

Reverse acquihires present a variation on that problem in emerging AI markets. A startup with limited present sales may ultimately fail, pivot, complement an incumbent, or grow into an independent competitor. Hiring its personnel or licensing its technology may accelerate innovation rather than diminish competition. But when the competitive significance of a transaction depends on which of those paths was realistically available, present market share alone may provide an incomplete picture. Section 7’s predictive framework permits the inquiry to encompass evidence bearing on those future competitive conditions while still requiring that predictions about nascent competition be grounded in the factual record.

Conclusion

Reverse acquihires demonstrate how the form of competitive consolidation can evolve even as the underlying questions posed by antitrust law remain familiar. In AI and other innovation-driven markets, a startup’s competitive significance may lie not only in its current products or market share, but also in the personnel, technology, and research trajectory from which future competition could emerge. Whether a particular reverse acquihire threatens that competition will depend on the structure of the transaction, the evidence surrounding the firms’ competitive relationship, and the probable trajectory of the startup had it remained independent.

For state antitrust enforcers, reverse acquihires may warrant attention as part of an evolving transactional landscape, in which competitive consolidation can occur outside the form of a traditional acquisition. Section 7 has long required courts and enforcers to assess probabilities and future competitive conditions before anticompetitive effects fully materialize. Reverse acquihires may be novel in transactional form, but they present a familiar Section 7 task: evaluating threats to competition before they fully emerge.

Endnotes

1. Kate Clark & Shirin Ghaffary, What Happens to AI Startups When Their Founders Jump Ship for Big Tech, Bloomberg (Aug. 4, 2025), https://www.bloomberg.com/news/articles/2025-08-04/what-happens-to-ai-startups-after-big-tech-lures-away-their-founders. [↩]
2. Krystal Hu & Harshita Mary Varghese, Microsoft Pays Inflection $650 Mln in Licensing Deal While Poaching Top Talents, Source Says, Reuters (Mar. 21, 2024), https://www.reuters.com/technology/microsoft-agreed-pay-inflection-650-mln-while-hiring-its-staff-information-2024-03-21/. [↩]
3. Clark & Ghaffary, supra note 1. [↩]
4. Id. [↩][↩][↩][↩][↩][↩]
5. Hu & Varghese, supra note 2 (including commentary from Professor Steven Weber of the UC Berkeley School of Information that “[t]his deal could be seen as an effort to reduce competition in the foundation model markets, as [Inflection] is going to be a shell of its former self”). [↩]
6. Sarah Bregel, What Is the Reverse-Acquihire?, Fast Company (Aug. 14, 2025), https://www.fastcompany.com/91384816/what-is-the-reverse-acquihire (“[M]ajor companies like Meta, Google, and Microsoft have been engaging in reverse-acquihires at AI startups[,] . . . swooping in to hire star talent and license technology [and] discarding the rest by the wayside.”). [↩]
7. An Update from Adept, Adept (June 28, 2024), https://www.adept.ai/blog/adept-update/ (announcing that Amazon would license Adept’s technology while Adept’s co-founders and part of its team joined Amazon’s AGI organization); Taylor Soper, Amazon Hires Founders from Well-Funded Enterprise AI Startup Adept to Boost Tech Giant’s “AGI” Team, GeekWire (June 28, 2024), https://www.geekwire.com/2024/amazon-hires-founders-from-well-funded-enterprise-ai-startup-adept-to-boost-tech-giants-agi-team/ (reporting that Amazon hired Adept’s co-founders and licensed portions of Adept’s technology while Adept continued operating independently). [↩]
8. Phoebe Liu, Sometimes You Don’t Want a GPU: Groq Cofounder Explains Whirlwind Deal With NVIDIA, Forbes (Mar. 18, 2026), https://www.forbes.com/sites/phoebeliu/2026/03/18/groq-cofounder-ross-explains-whirlwind-ai-chip-deal-with-nvidia/. [↩][↩]
9. Sherin Shibu, Meet Alexandr Wang, the 28-Year-Old Who Went From MIT Dropout to Billionaire Meta Hire: “I Wanted to Make a Difference”, Entrepreneur (June 16, 2025), https://www.entrepreneur.com/business-news/who-is-alexandr-wang-the-founder-of-scale-ai-joining-meta/493281. [↩]
10. The Wall Street Journal in June 2024 reported that the Federal Trade Commission was investigating whether Microsoft structured its deal with Inflection to avoid government antitrust review. Dave Michaels & Tom Dotan, FTC Opens Antitrust Probe of Microsoft AI Deal, Wall Street Journal (June 6, 2024). [↩]
11. See Clayton Act § 7, 15 U.S.C. § 18 (prohibiting an acquisition whereby “the effect of such acquisition may be substantially to lessen competition, or to tend to create a monopoly”); Brown Shoe Co. v. United States, 370 U.S. 294, 317-18 (1962) (describing that “Congress used the words ‘may be substantially to lessen competition’ . . . to indicate that its concern was with probabilities, not certainties,” and to arrest “tendencies toward concentration in industry” “in their incipiency”). [↩]
12. Jean-Michel Benkert, Igor Letina & Shuo Liu, Startup Acquisitions: Acquihires and Talent Hoarding, 178 Eur. Econ. Rev. 105103 (2025), https://www.sciencedirect.com/science/article/pii/S0014292125001539?via%3Dihub (discussing that startups “almost by definition, hold small or nonexistent market shares” and arguing that acquihires may function as preemptive strategies that distort innovation incentives and reduce consumer welfare even absent traditional market-share effects). [↩]
13. Clayton Act § 7, 15 U.S.C. § 18. [↩][↩]
14. United States v. Philadelphia National Bank, 374 U.S. 321, 362 (1963). [↩]
15. 370 U.S. 294, 317-18 (1962). [↩]
16. 374 U.S. at 362. [↩]
17. 370 U.S. at 323. [↩]
18. See United States v. General Dynamics Corp., 415 U.S. 486, 501-04 (1974). [↩]
19. See 908 F.2d 981, 982-83, 991-92 (D.C. Cir. 1990). [↩]
20. Id. (explaining that a prima facie case shifts the burden of production to defendants, but that “[t]he more compelling the prima facie case, the more evidence the defendant must present to rebut it successfully,” and crediting rebuttal evidence concerning entry, customer sophistication, bidding dynamics, and market-share volatility). [↩]
21. See 88 F.4th 1036, 1049-53 (5th Cir. 2023). [↩]
22. Id. at 1049-50, 1053. [↩]
23. See id. at 1052-53. [↩]
24. Id. at 1050. [↩]
25. Jeffrey Neal Johnson, NVIDIA’s $20B Groq Deal Is a Warning Shot to AI Rivals, MarketBeat (Dec. 26, 2025), https://www.marketbeat.com/originals/nvidias-20b-groq-deal-is-a-warning-shot-to-ai-rivals/. [↩]
26. Complaint ¶ 111, 114-15, New York v. Facebook, Inc. (D.D.C. Dec. 9, 2020) (No. 1:20-cv-03589). [↩]
27. Id. at ¶ 115. [↩]
28. Substitute Amended Complaint for Injunction and Other Equitable Relief at ¶¶ 14-15, FTC v. Facebook, Inc. (D.D.C. Sept. 8, 2021) (No. 1:20-cv-03590). [↩]
29. FTC v. Meta Platforms, ___ F. Sup. 3d ___, 2025 U.S. Dist. Lexis at *28-29, 130 (D.D.C. Nov. 18, 2025). [↩]