A familiar warning now shapes much of the discussion about artificial intelligence: A handful of dominant firms will control the technologies, stifle innovation, and require aggressive antitrust intervention. It is a compelling story—and mostly wrong.
The idea that large companies automatically mean less innovation has become conventional wisdom in antitrust circles. European regulators have embraced it, blocking mergers and attacking American tech companies. The Biden administration followed that path, treating size itself as a threat and wanting government-led AI. The Trump administration, by contrast, has signaled a more evidence-based view—one grounded in both economic logic and empirical studies.
The intellectual case against large firms innovating is not new. It traces at least to Kenneth Arrow’s 1960s paper holding that monopolists are less likely to innovate because they have much to lose when disrupting their own products. Later authors added concerns about stifling bureaucracy, groupthink, and excessive focus on existing customers. All these theories rest on an older idea—the structure-conduct-performance paradigm—which assumes that the number of firms determines how markets behave.
But that framework gets causality backward. As Joseph Schumpeter explained decades earlier, firms innovate precisely to reshape markets. Innovation is not a consequence of market structure—it creates it. Amazon didn’t enter markets to be like everyone else; it entered because high margins signaled opportunity. In dynamic industries like AI, today’s market is the byproduct of yesterday’s innovations—and the starting point for tomorrow’s.
There is also a deeper inconsistency in the “big mean less innovation” view. Counting firms assumes they are apples-to-apples comparable. But innovation is the process by which firms make themselves different. It’s incoherent to simultaneously treat firms as interchangeable for measurement and then credit them with meaningful differentiation.
And if the big-is-bad theory is clearly flawed, the empirical record is even more revealing.
Measuring innovation is inherently difficult. Studies tend to use two measures: patents and R&D. Patents are an imperfect proxy: Not all innovations are patented, and not all patents matter. R&D spending is equally blunt, varying widely in quality and purpose.
Yet despite these challenges, numerous studies have tried to see if size impacts innovation. I’ll focus on studies of mergers. The evidence does not show a systematic suppression of innovation from mergers. Indeed, the opposite is generally true—particularly in technology sectors.
Some cross-industry studies find that R&D intensity declines after mergers—but largely because sales rise, not because research falls. Research on so-called “killer acquisitions” sometimes observes reduced patenting after acquisitions, but these declines reflect a simple reality: Some startups patent aggressively to attract takeovers, and that behavior disappears once they are acquired.
In pharmaceutical mergers, results are varied. Studies that treat mergers as exogenous events—i.e., analyses that ignore why mergers occur—can find reduced R&D. But once researchers account for the reality that firms choose to merge—perhaps to acquire capabilities or refocus research pipelines—the reduction disappears.
Technology sectors tell a clearer story. In industries such as hard drives and cloud computing, mergers often coincide with increased R&D investment or improved innovation outcomes. Studies of acquisitions by large tech firms find that acquired technologies frequently become more widely used and cited after acquisition—evidence of broader diffusion, not suppression.
Other findings point to a “race to be acquired,” particularly by the GAFAM tech companies—Google (Alphabet), Apple, Facebook (Meta), Amazon, and Microsoft. Startups increase innovation efforts in hopes of being purchased by major platforms. Acquisitions sometimes lead rivals to scale back—not because innovation is stifled, but because exit pathways have changed. Venture capital patterns reinforce this: Investment shifts toward areas with active acquisition markets, suggesting that mergers can stimulate, rather than deter, entrepreneurial activity.
To be sure, not every merger enhances innovation. Outcomes vary depending on technologies and firms’ capabilities. But the blanket claim that large successful businesses harm innovation finds little support outside stylized theory and confirmation bias.
The implication is straightforward. Regulators should not presume that large firms—especially in fast-moving sectors like AI—threaten innovation. Indeed, the burden of proof should rest on those seeking to block deals, not on firms attempting to combine resources to innovate.
Large firms are not slowing AI; naïve regulatory policies do.