Stacey Dogan writes that antitrust regulators in the United States and Europe are right to investigate Big Tech-AI partnerships. Even if AI markets remain competitive today, history and economics show that the Big Tech companies will push to monopolize segments of the AI market if given the opportunity. The investigations serve as a deterrent against anticompetitive behavior and give the regulators access to the knowledge and information that will be necessary to detect anticompetitive patterns as the AI market matures.
Editor’s Note: This article is part of a symposium which asks experts to evaluate the anticompetitive harms of Big Tech investments in AI startups in light of recent investigations from antitrust agencies on both sides of the Atlantic. See here to read the other contributions from Matt Perault, Vivek Ghosal, and John Kirkwood.
Antitrust authorities in the United States and around the world have expressed keen interest in the growing web of relationships between dominant tech firms and artificial intelligence (AI) developers. The relationships have come fast and furiously, accompanied by hefty price tags and dressed up in a variety of transactional forms. The firms present the deals as innovation and efficiency-enhancing and characterize the AI marketplace as highly competitive.
Regulators appear unconvinced. While we have yet to see a formal challenge to any of these deals, competition authorities in the U.S., European Union, United Kingdom, and elsewhere have taken a position of active vigilance. Through information requests, investigations, opinions, policy statements, and public comments, regulators have made clear that they will not sit on the sidelines as the firms jockey for dominant positions in these evolving markets. Instead, regulators will evaluate deals as they happen, with the goal of responding nimbly to those that are likely to have anticompetitive effects.
The strategy of active vigilance reflects a modest, pragmatic attempt to preserve the disruptive potential of AI technologies and to promote competition in AI markets. It also reflects a desire to learn from the failure of the wait-and-see approach that regulators took in the last rounds of tech disruption, such as personal computing, search, and social media.
Disruptive technologies like AI have the potential to upend existing markets, dislodge dominant players, and inspire future disruptors. As described by economist Joseph Schumpeter, these “gales of creative destruction” keep market leaders on their toes and reward upstarts who supplant them in the hope of achieving their own (temporary) dominance.
Disruption can bring a wide range of social benefits, from technological innovation to improved efficiency to better products and lower prices for consumers. To the extent that it reallocates power in the broader economy, disruption also holds the potential to avert plutocracy, by loosening the grip that a small group of entrenched players can obtain over our economy, politics, and society.
Even without the barrage of recent deals, the characteristics of AI make the prospect of outsider disruption elusive, particularly for foundation models due to the amount of computing power and data needed to train them. Some of the problem lies with the nature of the technology itself: the barriers to entry are daunting, and participation in the market requires access to resources—such as specialized chips, computers, data, talent, and capital—that are controlled (sometimes uniquely) by Big Tech firms like Google and Amazon. Many of these same firms, moreover, control access to the platforms best suited to deliver certain AI applications to consumers such as search engines and social media networks.
As a result, the vision of an AI revolution spearheaded by independent outsiders who sweep aside powerful tech incumbents appears farfetched. At least in the near term, existing tech powerhouses will inevitably play a significant role in the creation and distribution of many AI-related products and services. Yet the inevitability of their participation need not lead to these firms’ dominance of AI markets; indeed, AI-related technologies could grow into healthy, competitive markets while simultaneously supplanting products (and displacing monopolists) in existing markets such as general internet search.
Even so, we can assume that today’s large firms will do everything they can to control new markets and to maintain control over existing ones, and experience has shown that tech monopolization can be hard to cure. As a result, competition authorities are wise to scrutinize the shape and composition of AI markets as they are developing and to intervene promptly if behavior is reducing competition or supporting a shift toward—or maintenance of—monopoly in any market.
Although the dizzying pace and diversity of AI development may make the current case for intervention look weak, these same features point to a marketplace in transition. Such transitional moments raise two primary competition-related risks: first, that incumbents whose existing market power is threatened by the disruption will either block the disruption or harness it to reinforce their dominance in a current market; and second, that a new market will, over time, become concentrated or monopolistic through anticompetitive agreements or exclusionary conduct rather than competition on the merits.
These risks are not hypothetical; leading tech firms have used each of these strategies to achieve or maintain their dominance in particular markets. In the 1980s, Microsoft tried to squash Netscape Navigator because of the perceived threat that the disruptive technology posed to its operating system monopoly. Facebook took a different tack, acquiring potential competitors to preserve its dominance in social media. Google, on the other hand, began as a disruptor but became a monopolist that has used exclusionary behavior to maintain its monopoly in online search.
While these threats may not play out in the AI environment, we can assume that the economic incentives to monopolize would motivate tech giants to return to their playbook if given the chance. In a recent joint statement, competition authorities from the U.S., EU, and U.K. discussed these risks as plausible in AI markets, given the growing investments by large tech firms in AI markets and their ubiquity throughout the AI stack. They flagged the possibility that firms might use control over inputs and distribution channels to limit competition in foundation-model markets or to thwart disruption of markets that they currently control. The regulators were careful to frame their statement as a warning, not a condemnation or a conclusion that anti-competitive behavior had yet occurred. But the statement provided notice that the enforcers are watching.
The message of regulatory vigilance is reinforced by the recent spate of investigations into deals and partnerships in the AI space. I review some of these investigations below.
Investments and partnerships in foundation model markets, especially Microsoft/OpenAI
Several investigations focus on investments by major tech firms in the foundation model market. In January 2024, the Federal Trade Commission announced an inquiry into “recent investments and partnerships” between leading tech firms and AI and cloud services providers, and recent reports suggest that the agency is especially focused on Microsoft’s $13 billion investment in OpenAI. Regulators in the EU and the U.K. are also scrutinizing the firms’ relationship.
These investigations raise two related questions. First, should the antitrust agencies treat the transaction as a merger, which would subject it to filing requirements and rigorous review of its impact on market concentration and on Microsoft’s overall share of the market? Second, if the transaction does not qualify as a merger, does it nonetheless have substantial anticompetitive effects in the foundation model market or related markets? To date, the EU competition regulator decided not to treat the deal as a merger but indicated a continuing investigation into potential anticompetitive effects, while the U.S. and U.K. have yet to resolve either question.
As the FTC’s specific inquiries show, the agencies are especially interested in whether the deal has elements that give control to Microsoft or exclusivity that might limit competitors’ access to one or more markets. If, despite its “minority economic interest,” Microsoft exerted control over OpenAI’s operations, it could potentially leverage that control to limit competition across multiple markets. Such control would likely inhibit competition between OpenAI and Microsoft’s own foundation models, and Microsoft could also frustrate competitors’ access to GPT products in sectors such as workplace software and search.
Conversely, a commitment by Microsoft to use OpenAI products exclusively or to grant exclusive access to Microsoft’s cloud computing service could foreclose important markets for other foundation model developers. The existence and scale of such effects depend on the overall competitive marketplace in addition to the terms of the deal itself. The agencies, however, are watching not only for realized harms, but for trends that might cause competitive harms in the future.
Microsoft and OpenAI, meanwhile, have gone to great lengths to prove OpenAI’s independence and to portray it as a competitor with Microsoft’s own AI products. Microsoft recently gave up its non-voting, observer position on OpenAI’s board, and a recent corporate filing described OpenAI as a competitor in both AI and internet search. The investigators will review written agreements, planning documents, and internal communications to evaluate whether the relationship between the parties creates de facto restrictions, even in the absence of any formal restraint. Similar investigations—and similar issues—have arisen over Amazon’s multimillion dollar partnership with Anthropic and other major investments between tech giants and AI firms.
Licenses that might foreclose access to inputs or outputs
In addition to investments and partnerships, the antitrust agencies are also showing interest in discrete deals that might impair competition in AI markets or preserve existing monopolies. The EU, for example, launched an investigation into Google’s exclusive agreement with Samsung to pre-install Google’s small model “Gemini nano” on Samsung’s Galaxy S24 series smartphones. While this deal alone may not have widespread exclusionary effects, it resembles the type of exclusive arrangements through which Google has preserved its dominance in online search. If Google managed to entrench its AI products on a substantial share of consumer devices, it could frustrate disruption of its search monopoly.
Nvidia
The U.S. Department of Justice and the French competition authority have both launched investigations into Nvidia, which manufactures the vast majority of the specialized chips (GPUs) that are essential to building foundation models. These inquiries are looking into whether Nvidia is maintaining its power in the GPU market through exclusionary acts, such as threatening to punish AI customers who buy chips from its competitors. Given Nvidia’s over-80% share of the GPU market, such exclusionary conduct, if it’s occurring, would likely justify antitrust action.
The “all-but-acqui-hires”
As the antitrust authorities have intensified their focus on AI-related investments and mergers, firms have invented new ways to access two critical inputs—technology and talent—in deals structured to avoid merger review or regulatory scrutiny. Amazon, for example, hired the co-founders and “close to” 66% of the employees of AI startup Adept, and obtained a non-exclusive license to use its technology. Microsoft hired the team and licensed the technology behind Inflection AI, a foundation model and consumer chatbot startup, which in the wake of the deal will shift its business to selling models to enterprise customers instead of providing its chatbot directly to nonbusiness users. Google reportedly hired Character.AI’s founders and obtained a non-exclusive license to its chatbot technology in a deal worth $2.5 billion.
While so-called “acqui-hires”—acquisitions made with the primary goal of obtaining a target firm’s talent—have occurred for many years, the practice of buying talent and technology without a formal acquisition emerged only in recent months, and is a patent attempt to circumvent merger rules. The FTC and the U.K. competition authority have announced investigations into these deals. As they do so, the agencies should view the transactions functionally rather than formally and consider whether any of the deals effectively snuffed out the target’s competitive position in the AI market.
As the diversity and size of these transactions demonstrate, the AI market is experiencing significant restructuring, with major tech firms expanding their footprints through investments, partnerships, quasi-acquisitions, and licensing deals. The frenetic jostling makes it difficult to assess market conditions and the likely competitive impact of any particular deal.
Even so, the current investigations promote three important regulatory goals. First, they will provide regulators with important information about how competition in these markets is evolving. Regulators can do their jobs more effectively if they have a sophisticated understanding of the state of these markets, including trends toward concentration, control relationships, cross-ownership, and exclusive licenses involving dominant tech firms. Second, the investigations may deter anticompetitive conduct by putting firms on notice that the regulators are prepared to intervene. This may be what occurred when Microsoft relinquished its non-voting, observer position on OpenAI’s board. Finally, the authorities are equipping themselves with the information necessary to act promptly when they identify anticompetitive conduct, before its effects become entrenched or irreversible.
Author Disclosure: the author reports no conflicts of interest. You can read our disclosure policy here.
Articles represent the opinions of their writers, not necessarily those of the University of Chicago, the Booth School of Business, or its faculty.
