Public Knowledge submitted an amicus brief recently opposing Google’s motion for a partial stay of the remedies in the search monopoly case. You can read the full brief here.
Search is how most people navigate the internet. It is where they start when they want to buy something, find a doctor, look up a news story, or settle an argument. And for most of the past two decades, search has meant Google. While it’s faced some challenges in recent years, Google Search is genuinely a great product, which is how it initially achieved its dominant position in the market.
The antitrust case against Google does not dispute that. What is in dispute is how Google kept its place in the market. Building a great product gets a company to the top. But paying billions of dollars a year to ensure that competitors can never reach users is something different. That is what this case is about.
The Department of Justice and a coalition of states sued Google in October 2020 under Section 2 of the Sherman Act, which prohibits companies from illegally maintaining monopoly power. After a full liability trial, Judge Amit Mehta found in August 2024 that Google had done exactly that. The court found that Google maintained its monopoly through a web of exclusive agreements with browser developers (primarily Apple and Mozilla) and with Android device manufacturers and wireless carriers, all of which locked Google into the default search position across virtually every device used in the United States. Google paid billions of dollars a year under these agreements, and the court concluded that these arrangements prevented rivals from getting off the ground.
A separate remedies trial concluded in May 2025, and the court entered a Final Judgment in December 2025. The remedies require Google to stop conditioning its distribution deals on exclusivity, share certain index data with qualified competitors, and make syndication (access to its results) available on fair terms. Google has now asked the court to pause most of the remedies while it appeals, which is why Public Knowledge submitted an amicus brief arguing the court should deny that request.
What the monopoly actually costs people
Google asks the court to treat the remedies as a greater problem than the conduct they address. But they’re not. The court found that Google’s conduct has harmed search users and the broader marketplace.
Choice is one area of harm. The court found that many users do not know what a default search engine is, let alone how to change one, and those who try face what the court called “choice friction” – multiple steps and confirmation screens that make switching annoying enough that most people give up. The court acknowledged that Google’s quality is part of why partners chose it as a default, but quality alone cannot explain a market where Google’s share sat at nearly 80 percent in 2009, grew to nearly 90 percent by 2020, and where its biggest rival, Bing, has never exceeded 11 percent and now sits below 6 percent. According to expert testimony in the trial, search engines are “experience goods,” meaning users have to actually try one to evaluate it, and the exclusive default agreements ensured that most users never had that experience with anyone but Google. One study the court considered offered users a small payment to try Bing for two weeks. A third of them kept it when the trial ended, and nearly two-thirds said it performed better than they had expected.
The financial harms run alongside the choice harms. Google used its dominant position to raise ad prices without any competitive constraint. The court found that Google deployed “pricing knobs” to inflate text ad costs in 5 to 15 percent increments, and deliberately designed those increases to look like normal auction fluctuations so advertisers would attribute rising costs to the market rather than to Google’s own choices. Businesses pass those costs on to consumers.
There is also the question of what monopoly power does to a company’s incentives over time. In 2020, Google ran an internal experiment that degraded major ranking components for about three months. The degradation was massive: equivalent, the court found, to removing twice the information contained in all of Wikipedia. But revenue declined by less than one percent. The point is not that Google produces “bad” results. The point is that a monopolist insulated from competition has less reason to invest in quality than one that faces a genuine threat of losing users, and that internal experiment confirmed exactly that dynamic.
Why a stay would be particularly damaging right now
Google’s request to pause the remedies frames the issue as something of a procedural matter: a brief pause while courts work through a complex appeal. But appeals in antitrust cases of this complexity take years, and the explosive growth of artificial intelligence (AI), much of which happened while the trial was still ongoing, will not wait. AI-powered search and information tools are posing the first real challenge to Google in years – but with its advantages in data, infrastructure, and reach into people’s lives, Google might be able to further extend its monopoly. As Cloudflare has documented, Google’s combined search and AI crawler lets it see 3.2 times more of the web than OpenAI and 4.8 times more than Microsoft, because publishers cannot block Google’s AI data collection without also disappearing from search results.
The court’s remedies opinion explicitly recognized that Google might try to extend its search distribution model into AI, and it crafted the remedies in part to prevent that from happening. As of the remedies trial, the AI market was still genuinely competitive: OpenAI held about 85 percent of U.S. users, with Google’s Gemini at 7 percent, Claude at 3 percent, and Perplexity and Copilot splitting the rest. Those numbers describe a market where positions have not yet solidified. Google is moving aggressively to lock up distribution before that window closes. It has already signed Gemini placement deals with Samsung, Motorola, and Lenovo, replicating precisely the default-agreement strategy the court found illegal in search.
Then, four days before filing its notice of appeal, Google announced a multi-year deal with Apple under which Apple will pay Google about $1 billion a year for access to Gemini as the foundation for Siri and Apple Intelligence. For years, Google paid Apple billions for exclusive default search placement, and the court found that arrangement was central to how Google maintained its search monopoly. It is true that with the Siri deal, the money flows the other way: Apple pays Google for AI access and hands it a distribution channel into 2.5 billion devices.
But the net balance of payments still favors Apple, so in practice the payments for Gemini amount to a discount on what Google was already paying Apple for search distribution. And for this discount, a rounding error or couch-cushion money for both companies, Google gets distribution its rivals cannot match. No competing AI company can draw on an existing financial relationship with Apple to reach that scale.
Google search queries on the Safari browser declined for the first time in 22 years last year, a sign that the transition to AI-powered search is already reshaping the market. Google’s AI Overviews feature is already recapturing some of that lost traffic. If the stay is granted and the appeal runs two or three years, Google may lock up AI distribution as thoroughly as it locked up search distribution.
The public has waited long enough
This case has been pending since October 2020. The liability trial concluded in November 2023. The remedies trial concluded in May 2025. The Final Judgment was entered in December 2025. The DOJ and state attorneys general devoted years and enormous resources to proving these violations, and they prevailed in the most significant antitrust enforcement action against a technology company in a generation. The public has already waited five years for the conduct to stop. The remedies are not drastic. Google has to modify some contracts to remove exclusivity provisions, build infrastructure to share index data, make syndication available on fair terms, and cooperate with a Technical Committee. For a company valued at over four trillion dollars with thousands of engineers, these are easy compliance obligations, not existential ones. Granting a stay would extend competitive harms that have persisted for over a decade, give Google more time to replicate its search distribution strategy in AI, and preserve a search market shaped more by exclusive dealing than by genuine competition. The court got it right, and it should keep its remedies in place.