Reverse Acqui-Hires and the New AI Exit: When Hiring the Team Starts to Look Like Buying the Company

I. Overview

A traditional acquisition is usually easy to recognize: a buyer purchases the target’s stock, acquires its assets, or merges with the target.[1] Artificial intelligence ("AI") dealmaking is making that picture less tidy because buyers can obtain a startup’s people, know-how, and technology rights without purchasing the startup itself.[2] Morgan Lewis reported that AI attracted more than half of global venture capital funding in 2025.[3] At the same time, "acqui-hires" and competition for AI talent reemerged, while diligence surrounding AI transactions became increasingly complex.[4] That matters for practicing lawyers because the economically important pieces of the transaction may be spread across employment agreements, technology licenses, and payments that look ordinary in isolation.[5] Reverse acqui-hires therefore raise a practical question first: what did the buyer actually obtain?[6]

II. From the Acqui-Hire to the Reverse Acqui-Hire

Traditional acqui-hires involve a buyer purchasing a startup principally to obtain its founders, engineers, or other employees, rather than to continue operating the startup’s existing business.[7] A reverse acqui-hire flips that structure: instead of buying the startup to get the team, the buyer hires the founders or key personnel directly, and separately obtains access to the startup’s technology, often through a non-exclusive license, while the target remains a separate legal entity.[8] For counsel, several coordinated agreements may therefore need to be evaluated together.[9]

Microsoft’s 2024 arrangement with Inflection AI ("Inflection") illustrates the structure.[10]  Microsoft hired Inflection co-founders Mustafa Suleyman and Karen Simonyan, together with almost all of Inflection’s team, and entered into associated arrangements that included a non-exclusive license to Inflection’s intellectual property.[11] But Microsoft did not purchase Inflection’s stock, and Inflection remained legally independent.[12] Similar structures later appeared at Adept, Character.AI, Windsurf, and Groq, pairing employee hiring with non-exclusive technology rights or related consideration.[13]

The accounting treatment of Character.AI and Groq helps explain why lawyers should not dismiss the structure as mere recruiting.[14] Alphabet recorded $2.7 billion of goodwill and $413 million of intangible assets in connection with Character.AI, while NVIDIA recorded $14.4 billion of goodwill and $2.5 billion of developed-technology intangibles in connection with Groq.[15] NVIDIA attributed goodwill primarily to the workforce and the expected future development of the licensed technology.[16] Those accounting entries do not determine whether either transaction constitutes an "acquisition" under federal antitrust law.[17] They do, however, show where the companies themselves placed value: in the people, technology, and future capabilities they obtained without purchasing the target’s equity.[18] Counsel on both sides of the deal should therefore ask what competitive capability moved and what remains with the target.[19]

III. The Microsoft-Inflection Test Case

The United Kingdom’s Competition and Markets Authority ("CMA") confronted that problem directly in its review of Microsoft and Inflection.[20] Instead of asking only whether Microsoft acquired Inflection’s stock or exclusive ownership of its technology, the CMA examined whether employees, collective know-how, intellectual-property access, and other arrangements together transferred enough of Inflection’s pre-transaction capabilities to create economic continuity.[21]

The CMA emphasized that almost all of Inflection’s team had moved to Microsoft and that the team brought collective know-how, established projects, priorities, and ways of working that  Microsoft could deploy immediately.[22] It therefore concluded that Microsoft had acquired an "enterprise" for purposes of U.K. merger jurisdiction, even though Inflection remained legally independent.[23] The CMA ultimately cleared the transaction after finding no realistic prospect of a substantial lessening of competition.[24]

U.S. antitrust law uses a different statutory framework, so the CMA’s jurisdictional conclusion cannot simply be imported into American law.[25] Its method is still useful for U.S.-based counsel because it asks what functions moved and what stayed behind.[26] Buyer-side diligence should extend beyond the license to the people, information, and operating capacity received; target-side counsel should ask whether the remaining company can continue competing.[27]

IV. Section 7 and HSR: Different Jobs, the Same Threshold Problem

Federal merger law presents two distinct but related questions.[28] Section 7 of the Clayton Act ("Section 7") supplies the substantive rule by prohibiting acquisitions of stock or assets when the effect "may be substantially to lessen competition, or to tend to create a monopoly."[29] The Hart-Scott-Rodino Antitrust Improvements Act ("HSR"), by contrast, generally requires parties to certain qualifying acquisitions to notify the federal antitrust agencies and observe a waiting period before closing.[30] HSR addresses advance notice and review of covered transactions, while Section 7 addresses whether an acquisition itself presents an unlawful competitive risk.[31] A transaction that does not require HSR filing is not, for that reason alone, insulated from later Section 7 scrutiny.[32]

Reverse acqui-hires make the relationship between those statutes harder because the threshold question is itself contested: has the buyer acquired stock or assets at all?[33] Buyer-side counsel may conclude that the hiring, license, and payment arrangements do not create an HSR-reportable acquisition, but that conclusion should not necessarily end the analysis if the same package transfers the target’s competitive capabilities.[34] Target-side counsel faces the parallel question whether the remaining entity still has the know-how, personnel, and operating capacity to compete.[35]

Recent scholarship approaches that problem from complementary directions.[36] David Wong has argued that Section 7 can reach reverse acqui-hires because a specialized team’s human capital, its knowledge, information, and skills, can constitute an intangible asset.[37] Nina Fridman focuses on HSR and argues that reverse acqui-hires can transfer intangible assets such as business information and know-how even when the buyer never purchases the startup itself.[38] Their statutory focus differs, but both theories reject the assumption that "assets" must mean only property transferred through a conventional acquisition agreement.[39] For practitioners, the combined insight is that separating personnel, technology, and consideration across documents may not control how antitrust law characterizes the overall transfer.[40]

V. When Hiring Becomes More than Hiring

Federal regulators have begun focusing on that boundary.[41] In January 2026, FTC Commissioner Mark Meador warned that some AI acqui-hires may be structured to avoid formal HSR review, and two months later emphasized a substance-over-form approach to digital market transactions.[42] The FTC and DOJ also sought public comment on nontraditional structures that may leave an acquired person legally intact but no longer competitively viable, expressly identifying licensing agreements combined with acqui-hires or reverse acqui-hires.[43] Reuters subsequently reported on September 10, 2026, that the DOJ was investigating whether NVIDIA structured its Groq licensing arrangement to avoid antitrust scrutiny.[44] That investigation is not a finding of illegality, but the sequence shows that transaction structure has become an enforcement issue.[45]

The line between hiring and acquiring still matters because ordinary employee mobility is not a merger.[46] Microsoft argued before the CMA that Inflection employees chose individually to resign and that employee "poaching" is an important feature of a free and competitive labor market.[47] The CMA did not treat hiring alone as an acquisition; it examined whether the broader arrangement transferred a functioning combination of people, know-how, and technology capable of continuing Inflection’s prior business activity.[48] A practical U.S. screen should consider the importance of the personnel who move, what technology and know-how rights accompany them, whether the agreements were negotiated as one package, how consideration flows, and what operating capacity remains with the target.[49] Functional significance can matter more than raw headcount because hiring twenty engineers from a large company is different from hiring nearly the entire team capable of developing a startup’s only product.[50]

That framework also exposes a harder question: falling outside premerger notification is not necessarily undesirable, so how broadly should these transactions be swept into premerger review?[51] A rule that treats every coordinated hiring and licensing arrangement as an HSR-reportable acquisition could sweep too broadly, burdening ordinary recruiting, technology licensing, and distressed-startup transactions that create useful "soft landings."[52] That concern is particularly important in the venture-backed startup ecosystem, where failure is far more common than a successful exit.[53] When a conventional acquisition is unavailable, a "soft landing" can allow a struggling startup to preserve some value through alternatives such as an acqui-hire or other transfer of its employees, technology, or assets.[54] Start-up failure scholarship emphasizes that these soft-landing arrangements can preserve technology, return value to stakeholders, and redeploy employees and capital when a full-company sale is a poor fit.[55] The agencies themselves acknowledge the tradeoff: their 2026 HSR inquiry asks how to address nontraditional transactions that may eliminate a market participant while reducing burdens on non-problematic deals.[56] The sharper concern is whether regulators receive a meaningful opportunity to review transactions that transfer a target’s competitive capabilities before integration occurs.[57]

That tradeoff should inform deal strategy on both sides.[58] Buyer-side counsel should identify early whether combined arrangements create filing or substantive antitrust risk even without a conventional acquisition agreement.[59] Target-side counsel may value speed and a soft landing, but it also should assess whether the transaction leaves a viable business behind and whether regulatory scrutiny could disrupt closing.[60] Morgan Lewis’ AI-deal analysis reinforces the broader point that data provenance, model-intellectual property, regulatory issues, and talent retention now require specialized diligence.[61] Better issue spotting comes from evaluating the package as a transaction rather than treating each workstream as self-contained.[62]

VI. Conclusion

Reverse acqui-hires expose a growing mismatch between the economic substance of some modern AI transactions and the legal forms around which merger review developed.[63] Antitrust law need not treat every employment offer or technology license as an acquisition, and premerger notification should not become a barrier to ordinary hiring or value-preserving soft landings.[64] But the analysis should not end merely because the buyer leaves the target's stock untouched and the employment, licensing, and consideration arrangements appear in separate agreements.[65] When those arrangements together transfer the people, know-how, technology rights, and practical business capability that made the target competitive, counsel on both sides should recognize that the transaction may present acquisition questions even though the target survives as a legal entity.[66] Part II turns from this public-law problem to the investors who financed the startup in the first place: what happens when the economic substance of an exit occurs without triggering the contractual protections tied to conventional exit events?[67]

[1] See David T. Wong, An Acquisition by Another Name: Reverse Acquihires Under the Clayton Act, 135 Yale L.J. 3175, 3179–91 (2026), https://yalelawjournal.org/comment/an-acquisition-by-another-name-reverse-acquihires-under-the-clayton-act; Nina Fridman, When a Mass Resignation Becomes a Merger: Rethinking Asset Acquisitions for the AI Era, 93 U. Chi. L. Rev. 999, 1011–16 (2026), https://chicagounbound.uchicago.edu/uclrev/vol93/iss4/4/.

[2] See Wong, supra note 1, at 3179–91; Fridman, supra note 1, at 1011–16.

[3] See Brian P. Slough & Andrew M. Ray, AI Deals in 2025: Key Trends in M&A, Private Equity, and Venture Capital, Morgan Lewis (Sept. 29, 2025), https://www.morganlewis.com/pubs/2025/09/ai-deals-in-2025-key-trends-in-ma-private-equity-and-venture-capital.

[4] See id.

[5] See Slough & Ray, supra note 3; Wong, supra note 1, at 3179–91; Fridman, supra note 1, at 1011–16.

[6] See Wong, supra note 1, at 3179–91, 3192–3200; Fridman, supra note 1, at 1011–16, 1022–24.

[7] See John F. Coyle & Gregg D. Polsky, Acqui-Hiring, 63 Duke L.J. 281, 283–84, 293–301 (2013), https://scholarship.law.duke.edu/dlj/vol63/iss2/1/.

[8] See Wong, supra note 1, at 3179–91; Fridman, supra note 1, at 1011–16.

[9] See Slough & Ray, supra note 3; Wong, supra note 1, at 3179–91; Fridman, supra note 1, at 1011–16.

[10] See Competition & Mkts. Auth., ME 7103/24, Microsoft Corporation’s Hiring of Certain Former Employees of Inflection and Its Entry into Associated Arrangements with Inflection: Decision on Relevant Merger Situation and Substantial Lessening of Competition, ¶¶ 1–3 (Sept. 4, 2024), https://assets.publishing.service.gov.uk/media/6719ff5f549f63039436b3c8/__Full_text_decision__.pdf (full text published Oct. 24, 2024).

[11] See id. ¶¶ 1–3, 30–33.

[12] See id. ¶¶ 1–3.

[13] See Greg Bensinger & Krystal Hu, Amazon Lures Cofounders from Startup Adept to Bolster AI Efforts, Reuters (June 29, 2024), https://www.reuters.com/technology/amazon-hires-ai-startup-adepts-cofounders-join-its-ai-org-2024-06-28/; Alphabet Inc., Annual Report (Form 10-K), at 76–77 (Feb. 5, 2025), https://www.sec.gov/Archives/edgar/data/1652044/000165204425000014/goog-20241231.htm; Kenrick Cai, Krystal Hu & Kritika Lamba, Google Hires Windsurf Execs in $2.4 Billion Deal to Advance AI Coding Ambitions, Reuters (July 11, 2025), https://www.reuters.com/business/google-hires-windsurf-ceo-researchers-advance-ai-ambitions-2025-07-11/; NVIDIA Corp., Annual Report (Form 10-K), at 60 (Feb. 25, 2026), https://www.sec.gov/Archives/edgar/data/1045810/000104581026000021/nvda-20260125.htm.

[14] See Alphabet Inc., supra note 13, at 76–77; NVIDIA Corp., supra note 13, at 60; Wong, supra note 1, at 3192–3200.

[15] See Alphabet Inc., supra note 13, at 76–77; NVIDIA Corp., supra note 13, at 60.

[16] See NVIDIA Corp., supra note 13, at 60.

[17] See Wong, supra note 1, at 3192–3200; Fridman, supra note 1, at 1022–24.

[18] See Alphabet Inc., supra note 13, at 76–77; NVIDIA Corp., supra note 13, at 60.

[19] See Competition & Mkts. Auth., supra note 10, ¶¶ 94–108; Fridman, supra note 1, at 1033–36.

[20] See Competition & Mkts. Auth., supra note 10, ¶¶ 1–3, 82–108.

[21] See id. ¶¶ 87–89, 94–108.

[22] See id. ¶¶ 95–106.

[23] See id. ¶¶ 99–100, 103–08.

[24] See id. ¶¶ 1, 23, 255.

[25] See 15 U.S.C. §§ 18, 18a; Competition & Mkts. Auth., supra note 10, ¶¶ 82–108.

[26] See Competition & Mkts. Auth., supra note 10, ¶¶ 82, 94–108; Fridman, supra note 1, at 1033–36.

[27] See Competition & Mkts. Auth., supra note 10, ¶¶ 94–108; Fridman, supra note 1, at 1033–36; Slough & Ray, supra note 3.

[28] See 15 U.S.C. §§ 18, 18a.

[29] See 15 U.S.C. § 18.

[30] See 15 U.S.C. § 18a(a).

[31] See 15 U.S.C. §§ 18, 18a.

[32] See Marian Bruno, Avoidance Devices Won’t Avoid HSR Penalties, F.T.C. (Nov. 14, 2019), https://www.ftc.gov/enforcement/competition-matters/2019/11/avoidance-devices-wont-avoid-hsr-penalties.

[33] See Wong, supra note 1, at 3192–3200; Fridman, supra note 1, at 1022–24, 1033–36.

[34] See Bruno, supra note 32; Fed. Trade Comm’n & U.S. Dep’t of Just., Antitrust Div., Request for Public Comment Regarding Making Improvements to the Premerger Notification and Report Form 2–3, 6 (Mar. 25, 2026), https://www.ftc.gov/system/files/ftc_gov/pdf/2026.03.25-HSR-RFI.pdf [hereinafter Request for Public Comment].

[35] See Competition & Mkts. Auth., supra note 10, ¶¶ 94–108; Fridman, supra note 1, at 1033–36.

[36] See Wong, supra note 1, at 3192–3200; Fridman, supra note 1, at 1022–24, 1033–36.

[37] See Wong, supra note 1, at 3192–3200.

[38] See Fridman, supra note 1, at 1022–24, 1033–36.

[39] See Wong, supra note 1, at 3192–3200; Fridman, supra note 1, at 1022–24, 1033–36.

[40] See Request for Public Comment, supra note 34, at 2–3, 6; Mark R. Meador, Comm’r, Fed. Trade Comm’n, Keynote Address at The Tech Antitrust Conference, Concurrences 3–4 (Jan. 15, 2026), https://www.ftc.gov/system/files/ftc_gov/pdf/meador-concurrences-keynote.pdf [hereinafter Meador I].

[41] See Meador I, supra note 40, at 3–4; Mark R. Meador, Comm’r, Fed. Trade Comm’n, Antitrust for Digital Markets 4–5 (Mar. 23, 2026), https://www.ftc.gov/system/files/ftc_gov/pdf/Antitrust-for-Digital-Markets-Forum-Meador.pdf [hereinafter Meador II].

[42] See Meador I, supra note 40; Meador II, supra note 41.

[43] See Request for Public Comment, supra note 34, at 2–3, 6.

[44] See DOJ Probes Nvidia’s Licensing Deal with AI Startup Groq, NYT Reports, Reuters (Sept. 10, 2026), https://www.reuters.com/legal/litigation/us-doj-probes-nvidias-licensing-deal-with-ai-startup-groq-nyt-reports-2026-09-10/ [hereinafter DOJ Probes Nvidia].

[45] See DOJ Probes Nvidia, supra note 44; Request for Public Comment, supra note 34, at 2–3, 6.

[46] See Competition & Mkts. Auth., supra note 10, ¶¶ 84(b), 87–89.

[47] See id. ¶ 84(b).

[48] See id. ¶¶ 82, 87–108.

[49] See Competition & Mkts. Auth., supra note 10, ¶¶ 82, 94–108; Fridman, supra note 1, at 1033–36.

[50] See Competition & Mkts. Auth., supra note 10, ¶¶ 94–108; Fridman, supra note 1, at 1033–36.

[51] See Request for Public Comment, supra note 34, at 2–3, 6; Elizabeth Pollman, Startup Failure, 73 Duke L.J. 327, 356–59, 370–77 (2023), https://scholarship.law.duke.edu/dlj/vol73/iss2/2/.

[52] See Request for Public Comment, supra note 34, at 2–3, 6; Pollman, supra note 51, at 333, 356–59, 370–77, 383–85.

[53] See Pollman, supra note 51, at 329–30.

[54] See Pollman, supra note 51, at 356–59.

[55] See Pollman, supra note 51, at 356–59, 370–77.

[56] See Request for Public Comment, supra note 34, at 2–3, 6.

[57] See Request for Public Comment, supra note 34, at 2–3, 6; Meador II, supra note 41.

[58] See Slough & Ray, supra note 3; Bruno, supra note 32; Request for Public Comment, supra note 34, at 2–3, 6.

[59] See Bruno, supra note 32; Request for Public Comment, supra note 34, at 2–3, 6.

[60] See Pollman, supra note 51, at 356–59, 370–77; Competition & Mkts. Auth., supra note 10, ¶¶ 94–108; Request for Public Comment, supra note 34, at 2–3, 6.

[61] See Slough & Ray, supra note 3.

[62] See Slough & Ray, supra note 3; Competition & Mkts. Auth., supra note 10, ¶¶ 82, 94–108; Request for Public Comment, supra note 34, at 2–3, 6.

[63] See Wong, supra note 1, at 3192–3200; Fridman, supra note 1, at 1022–24, 1033–36; Request for Public Comment, supra note 34, at 2–3, 6.

[64] See Competition & Mkts. Auth., supra note 10, ¶ 84(b); Request for Public Comment, supra note 34, at 2–3, 6; Pollman, supra note 51, at 356–59, 370–77.

[65] See Competition & Mkts. Auth., supra note 10, ¶¶ 84(b), 87–108; Wong, supra note 1, at 3179–91, 3192–3200; Fridman, supra note 1, at 1011–16, 1022–24.

[66] See Competition & Mkts. Auth., supra note 10, ¶¶ 95–108; Wong, supra note 1, at 3192–3200; Fridman, supra note 1, at 1022–36.

[67] See Nat’l Venture Cap. Ass’n, NVCA Model Certificate of Incorporation (Updated Oct. 2025), https://nvca.org/document/nvca-model-certificate-of-incorporation-updated-oct-2025/ (last visited Oct. 3, 2026).