_Paper: Yangyang Cheng, D. Daniel Sokol, and Carmelo Cennamo, "Antitrust Enforcement and the Redirection of Platform M&A Strategy," Platform Strategy Symposium, 2026 | https://questromworld.bu.edu/platformstrategy/wp-content/uploads/sites/49/2026/06/1090_Cheng_Platform-MA-1.pdf
A premium, not a wall
Yangyang Cheng, D. Daniel Sokol, and Carmelo Cennamo built a dataset of 6,901 software sector M&A deals between 2000 and 2022 to ask a narrower, more useful question than "did antitrust enforcement deter acquisitions." They asked what kind of acquisitions it deterred. Their answer turns on a distinction they call the regulatory acquisition premium: the enforcement cost of a deal rises with how closely the target's capabilities overlap with the acquirer's own. A substitutive acquisition, where the target does roughly what the acquirer already does, looks to a regulator like a straightforward removal of a competitor. A complementary acquisition, where the target fills a capability gap the acquirer lacks, looks like ordinary portfolio building and draws less scrutiny.
Once enforcement intensified around 2017, the authors find that substitutive M&A among platform incumbents fell by 11.2 percent, concentrated among young targets, the companies most likely to have grown into real rivals. That sounds like deterrence working as intended. It is not the whole story. The same data show incumbents did not stop building out their ecosystems. They shifted the target type: away from ecosystem deals that would have extended a platform's own stack, and toward horizontal deals in adjacent lines of business, while vertical acquisitions were essentially untouched. Critically, the functional overlap between acquirer and target did not fall even as the substitutive deal count did, which the authors read as evidence that platforms substituted toward targets that are less scrutinized but functionally similar to the ones they gave up. The thing regulators wanted to stop, incumbents absorbing capability that looks like a competitor's, kept happening. It just stopped looking like the thing regulators were watching for.
That reframe matters because the natural way to evaluate antitrust policy is to count blocked deals and declare victory. Two of the most visible abandonments in recent memory illustrate exactly why that count is incomplete.
Two headline deaths, one quiet survivor
Adobe's attempt to buy the collaborative design platform Figma, a deal reported at roughly 20 billion dollars, collapsed on December 18, 2023. The UK's Competition and Markets Authority had opened a Phase 2 investigation in July 2023, published provisional competition findings that November, and set out a notice of possible remedies that would have required divesting core products. Adobe and Figma concluded, in their own words reported at the time, that there was no clear path to clearance in the UK or the European Union, and the companies terminated the agreement. The CMA's case file confirms it formally cancelled the merger reference on December 18, 2023. The same day, the Justice Department's top antitrust official, Assistant Attorney General Jonathan Kanter, issued a statement welcoming the outcome, saying the decision "ensures that designers, creators, and consumers continue to get the benefit of the rivalry between the two companies going forward."
A little over a month later, Amazon followed a similar script. Its proposed 1.4 billion dollar acquisition of Roomba maker iRobot was terminated on January 29, 2024, after European Union regulators signaled they would block it. The FTC's own statement, issued the following week, said the Commission's investigation into the deal "revealed significant concerns about the transaction's potential competitive effects," centered on Amazon's ability and incentive to favor its own products and disfavor rivals on its marketplace, with knock-on effects for innovation, entry barriers, and consumer privacy.
Both cases fit the Cheng, Sokol, and Cennamo framework cleanly. Figma competed directly with Adobe's own design tools; iRobot would have handed Amazon a dominant position in a product category the company could cross-sell and cross-promote on its own platform. Both were large, visible, substitutive in exactly the sense the paper defines, and both were abandoned. If you stopped the story there, antitrust enforcement looks like it is working on its own terms.
But look at what Microsoft did with Inflection AI in the same stretch of time, and the redirection half of the paper's argument becomes concrete rather than statistical.
The acquisition that declined to call itself one
In March 2024, Microsoft announced it was hiring most of the roughly 70 person staff of the AI startup Inflection, including co-founder Mustafa Suleyman, who became the head of a newly created Microsoft AI division. Reporting from Reuters and the Wall Street Journal, confirmed by CNBC, put the accompanying payment at 650 million dollars, structured largely as a licensing fee for Inflection's AI models so Microsoft could resell them through Azure, rather than as an acquisition of the company itself. Microsoft's own public position was blunt: a spokesperson told CNBC the company was "confident that the hiring of talent promotes competition and should not be treated as a merger."
The UK's Competition and Markets Authority disagreed with the framing, if not with the outcome. It opened a formal merger inquiry on July 16, 2024, specifically to decide whether the hiring, combined with "associated arrangements" with Inflection, amounted to a relevant merger situation under the Enterprise Act. On September 4, 2024, the CMA cleared the deal, but its decision recorded that the jurisdictional threshold for a merger had in fact been met, based on Microsoft and Inflection's combined and overlapping share of chatbot supply in the UK, before concluding there was no realistic prospect of a substantial lessening of competition. In substance, the regulator treated this as a merger. In form, no merger was ever filed for review in the way a share purchase would have been.
This is the regulatory acquisition premium playing out in real time, not in a dataset years later. Inflection's chatbot technology was about as close to Microsoft's existing Copilot business as Figma's design tools were to Adobe's, which is to say, substitutive by the paper's own logic. A straightforward acquisition of Inflection would have invited the scrutiny that killed Figma and iRobot. What Microsoft got instead, the technology, the talent, and effective control of the capability, through a licensing payment and a hiring wave, was a complementary looking transaction wrapped around a substitutive outcome. The target didn't change in function. The deal structure did.
What the guidelines actually targeted, and what they missed
None of this happened in a policy vacuum. On December 18, 2023, the same day the Adobe-Figma merger collapsed, the FTC and the Justice Department jointly released the final 2023 Merger Guidelines, the first full revision since 2020. The agencies' own press release described the guidelines as addressing competition "ranging from price competition to competition for the terms and conditions of employment, to platform competition," naming platform dynamics as a distinct analytical lens for the first time in the guidelines' history. FTC Chair Lina Khan framed the effort as reflecting "the new realities of how firms do business in the modern economy," and the guidelines also flagged that a series of smaller acquisitions could be examined together rather than deal by deal, a direct answer to the roll-up strategy of accumulating small rivals one quiet purchase at a time.
That provision is the right instinct, and the Cheng, Sokol, and Cennamo findings suggest it is not yet biting hard enough. Enforcement that scores success by the size and visibility of the deals it blocks will keep missing the deals it never gets a chance to review, because platforms have every incentive to make the next acquisition look nothing like the last one that got killed.
A deal that does not show up as a merger does not disappear. It shows up somewhere else on the balance sheet.
For managers inside platform corporate development groups, the lesson runs the other direction from the policy one, and it is just as actionable. Treat the regulatory acquisition premium as a planning variable, not a surprise. A target that duplicates an existing product line should be assumed to carry the highest review risk and the longest timeline, regardless of size, while a target extending into a genuinely new capability will clear faster. When a substitutive target is strategically essential anyway, as Inflection's technology clearly was for Microsoft's Copilot ambitions, the precedent shows there is a menu of structures, a non-exclusive license, a talent hire, a minority stake, that can deliver most of the strategic value without tripping the formal merger wire, though each one sacrifices something: less integration certainty, weaker intellectual property control, or a regulator who, as the CMA did here, decides to look anyway and simply takes longer to clear it.
The headline count of blocked megadeals will keep falling as enforcement gets more visible and more aggressive. Total consolidation, on the evidence from 6,901 deals and the quieter experience of Inflection AI, is not falling with it. It is just getting harder to see.
Sources
- Yangyang Cheng, D. Daniel Sokol, and Carmelo Cennamo, "Antitrust Enforcement and the Redirection of Platform M&A Strategy," Platform Strategy Symposium, 2026 questromworld.bu.edu
- 2026 Platform Strategy Symposium conference program (confirms the paper's listing, authors, and session), Boston University Questrom questromworld.bu.edu
- "Federal Trade Commission and Justice Department Release 2023 Merger Guidelines," Federal Trade Commission ftc.gov
- "Antitrust AAG Kanter Statement After Adobe and Figma Abandon Merger," U.S. Department of Justice justice.gov
- "Adobe / Figma merger inquiry" case page, UK Competition and Markets Authority gov.uk
- "Statement Regarding the Termination of Amazon's Proposed Acquisition of iRobot," Federal Trade Commission ftc.gov
- "Microsoft / Inflection inquiry" case page, UK Competition and Markets Authority gov.uk
- "Microsoft's hiring of staff from AI startup Inflection referred for UK merger probe," CNBC cnbc.com