AI Monopolies Are Coming. Now’s the Time to Stop Them.

Nine months after a federal judge found Google guilty of operating an illegal search monopoly, the tech giant is back in court sparring with antitrust regulators over how it should be punished. The trial, which began April 21, has presented sharply divergent paths for the company’s future: the Justice Department urged the court to force Google to sell its Chrome browser and share valuable data with rivals, while Google pushed for much lighter interventions.

The company’s fate will ultimately rest with Judge Amit Mehta, who is expected to decide on a penalty by the end of the summer. While his ruling will ostensibly focus on Google’s dominance in search, it could also impact the company’s ambitions in another important sector, artificial intelligence. If forced to share search data and divest from Chrome, Google could lose its grip on the levers that would let it extend its search monopoly into AI.

Though still in its early stages, the AI industry is already in danger of becoming monopolized by powerful tech companies like Google, which have cemented their dominance using many of the same tactics that they used to dominate the internet economy. Allowing these tech giants to capture the US AI ecosystem will make this critical new sector less innovative, less resilient, and less responsive to government policy, creating serious risks for the American people and national security community. By approving the government’s proposed remedies, Judge Mehta could help foster a more innovative, resilient, and responsive AI market, one in which entrepreneurs and technologists succeed on the strength of their ideas rather than their allegiance to Big Tech.

Since the first generative AI tools came online in late 2022, Google and other powerful tech companies like Amazon, Meta, and Microsoft have steadily tightened their grip over the nascent industry. Using the expansive troves of cash and technical assets amassed through their other businesses, the tech giants have vaulted to the forefront of AI development and positioned themselves as essential suppliers within the tech ecosystem. Today, it is virtually impossible for a startup to build, deploy, and distribute an AI product without relying, at least in part, on resources provided by Big Tech.

The tech giants are already exploiting this market power to their advantage. The companies have leveraged their computing infrastructure to gain stakes in promising labs like OpenAI and Anthropic, while absorbing other startups outright through traditional acquisitions and so-called “acquihires.” Even ostensibly independent developers can still be squeezed by the tech giants, who often control access to compute infrastructure, foundation models, training data, and other key ingredients for AI development.

As AI adoption accelerates, incumbent tech companies will also be able to leverage distribution platforms to give themselves and their affiliates a leg up on competitors. AI startups cannot survive without users, and today, many of the most popular channels for reaching those users—mobile devices, online marketplaces, software packages, web browsers—are controlled by incumbent tech companies. As I detail in a new report from Georgetown University’s Center for Security and Emerging Technology, the tech giants have ample opportunities to use their platforms to block rivals from gaining a foothold in the market. Indeed, the Justice Department’s proposed breakup of Google is intended, in part, to prevent precisely this type of exclusion.

Such behavior would hardly represent a new strategy—companies like Google, Amazon, Meta, and Microsoft have long relied on acquisitions, bundling, and self-preferencing to dominate their respective corners of the digital economy, as recent antitrust investigations have revealed. Letting them rerun this playbook in the AI industry would represent a major policy failure, leaving the sector less innovative, less resilient, and less responsive to the American public.

Historically, competition has been a key driver of technological innovation, pushing entrepreneurs to continually improve their offerings to get ahead of rivals. The idea that a scrappy startup can overtake a staid incumbent with an ingenious invention is a core tenet of American capitalism—and the mythos of Silicon Valley itself. Allowing incumbent tech companies to stifle competition in the AI ecosystem would undermine this disruption process, depriving the United States of breakthroughs necessary for maintaining global AI leadership.

For example, consider how the major American developers were caught off guard by DeepSeek, a Chinese company that, under pressure from US economic controls, was forced to innovate more efficient AI techniques. If the American AI ecosystem were not consolidated around a handful of enormously wealthy incumbents, such disruptions may have emerged domestically rather than across the Pacific.

Beyond stifling innovation, a concentrated AI ecosystem is also more vulnerable to supply chain disruptions, technological surprises, and other risks associated with digital monocultures. It could also make the industry more resistant to AI governance efforts, as powerful incumbents gain greater leeway to shape, block, or fight regulations.

This future is not inevitable, however. The federal government has a range of tools at its disposal to foster a more open, competitive, and diversified AI industry. One, of course, is antitrust enforcement. President Trump has, for now, proved willing to continue antitrust enforcement against Big Tech, and we may see ongoing lawsuits against Amazon, Apple, and Meta also grapple with AI market power.

To be sure, however, antitrust enforcement only cracks down on anticompetitive conduct after the fact. As such, fending off AI monopolies will also require policymakers to pursue more proactive measures to promote competition. These measures may include mandating interoperability among cloud providers, establishing a public cloud, restricting self-preferencing and other exclusionary conduct, enacting data portability requirements, and promoting the development of less resource-intensive models.

Such measures will likely provoke a strong backlash from Big Tech, and seeing them through will require considerable political courage. But the stakes are equally high. The choices that Judge Mehta and other policymakers make in the coming months and years will determine whether America’s AI industry will be locked down by yesterday’s tech giants or opened up to tomorrow’s innovators. If the country hopes to maintain its technological and economic edge, it must choose the latter.