Google Keeps AdX as Judge Orders Fixes to the Auction It Still Runs

Google will stop giving its own ad exchange first access to every impression before competitors can bid. It will stop seeing what rivals bid before it submits its own offer. It will stop forcing publishers to apply identical price floors across all buyers.

Those are the three mechanisms US District Judge Leonie Brinkema of the Eastern District of Virginia ordered overhauled on Tuesday, according to TechTimes, which reported that the ruling spares Google a forced sale of its ad exchange. Google keeps DoubleClick for Publishers (DFP) and AdX.

The April 2025 liability ruling that preceded Tuesday's decision found that Google had illegally monopolized two distinct markets under Sections 1 and 2 of the Sherman Act. The first was the publisher ad server market, where it held roughly 91% of worldwide market share as measured by impressions served. The second was the open-web display ad exchange market, where AdX had charged a consistent 20% take rate on every transaction for over a decade without competitive pressure to reduce it. Brinkema also found that Google had illegally tied the two together, using publishers' dependence on DFP to lock them into AdX, and using advertiser demand on AdX to lock publishers into DFP. That finding was the foundation. Tuesday's ruling was about the cure.

What the DOJ asked for

The Department of Justice had asked Brinkema to order sale of AdX within twelve months, the open-sourcing of DFP's final auction logic under a neutral administrator such as Prebid.org within twelve months, and a contingent divestiture of remaining ad server components if market competition did not recover within three years. According to TechTimes, government lawyers argued in closing that behavioral promises from a company whose anticompetitive conduct was "buried in computer codes and algorithms" could not be trusted. Only divestiture, they said, could "eradicate Google's illegally acquired monopolies root and branch."

Google countered that breaking up a system handling 55 million ad requests per second would be technically unworkable and economically harmful to the small publishers that use DFP at no charge. It proposed behavioral relief instead: real-time bid sharing, unified pricing rule removal, and compliance monitoring for six years. Brinkema sided with Google on structure and with a blended version of both parties' behavioral proposals on conduct. Her full written opinion remains sealed for fourteen days while both sides review it for confidential business information, but a short public order made the outcome clear. Both parties have thirty days to confer and file a jointly proposed final judgment.

First look, last look

The three prohibited practices map directly onto the specific conduct her liability ruling had documented.

First look, ended. Since approximately 2010, before any other exchange could bid on a publisher's ad impression, AdX received first access to acquire it at or above the publisher's floor price. Competing exchanges could only bid on impressions AdX had already declined. The Index Exchange antitrust complaint, filed in November 2025, describes this as giving AdX priority access since 2010. That is now prohibited.

Last look, ended. AdX also received the ability to see the highest bid submitted through rival exchanges before submitting its own competing offer. Brinkema's prior ruling, citing the Index Exchange complaint, described this as opening the sealed bid envelope before bidding. Rather than winning auctions on merit, AdX could simply outbid the current leader by one cent, every time, without risk. That is now prohibited.

Publisher price floors restored. In 2019, Google introduced what it called Unified Pricing Rules, which required publishers to apply the same minimum price threshold to every bidder, including AdX. What that change eliminated was publishers' ability to set higher minimum prices for AdX specifically, a lever that, in a competitive market, would counteract Google's structural advantages. Brinkema has now ordered those rules deprecated, restoring publishers' right to set different price floors for different buyers, as AdExchanger reporting confirms.

Real-time bid data to be shared. Going forward, AdX must make its bid amounts available to rival publisher ad servers in real time. This addresses the informational advantage that made AdX nearly mandatory for publishers seeking Google's buy-side advertising demand.

What did not change

What remains unchanged is the infrastructure that produced these practices. Google still owns the dominant publisher ad server and the dominant ad exchange. It still runs the auction software that decides which impression goes to which buyer. It still controls the data pipeline that tells both sides of the market how to price.

Jay Friedman, a co-founder of the ad-tech advisory firm CartographAI who testified for the DOJ at the remedies trial, put the residual problem as a question: "What is a web publisher to do if it wants to use a different ad server but still get Google's buy-side demand?" Real-time bid access is supposed to answer that, but whether it does will be determined in practice, not in the courtroom. Friedman also told AdExchanger that Brinkema had not adequately weighed the additional harm to publishers from Google's growing gene, a passage that was cut off in the TechTimes account.

For publishers, the practical effect is narrower than the headline. The ruling restores a pricing lever and opens the auction sequence, but it does not hand them an alternative to the vertically integrated stack they depend on for revenue. The behavioral measures run alongside compliance monitoring for six years, a period that outlasts the current administration but not the market structure the case was built around.

The wider antitrust file

Google's ad tech case is not the only antitrust matter moving through US regulators. The Verge reported on 1 June that the Federal Trade Commission is investigating Microsoft over potentially exclusionary behavior around its Azure cloud services and its role in the AI industry. The probe began under the Biden administration in 2024 and continued under President Donald Trump. The Verge obtained previously unreported information about civil investigative demands, which are similar to subpoenas, from an industry source who reviewed them and was granted anonymity to speak on nonpublic information. According to that source, the CIDs say the goal is to determine if Microsoft has used unfair methods of competition in its cloud, software products, and related services in violation of the FTC Act. Microsoft spokesperson Alex Haurek said the company is "cooperating fully with the FTC and believe our practices promote competition while delivering the innovative products our customers expect."

Customers have complained, often anonymously for fear of retaliation, that Microsoft's 2019 changes to its licensing terms made it significantly more costly to run Windows software on infrastructure outside of Microsoft's Azure cloud. In 2023, Google responded to a broad FTC inquiry about cloud computing by accusing Microsoft of using dominance in other areas to "give their own cloud products an unearned advantage" and lock in consumers. Haurek pointed to Google as "a clear example of that dynamic, growing 63% year over year and competing head-to-head with other major providers including AWS, the largest cloud provider."

There is no guarantee the investigation ends in a legal complaint. After further investigative steps, FTC staff will choose whether to recommend filing one, and the agency's two commissioners will take a vote. George Washington University law professor and former FTC Chair William Kovacic told The Verge that this kind of information gathering tends to occur at the earlier stages of an investigation. The two-member, all-Republican commission could choose to end it quietly without bringing a complaint, with no minority party commissioners to object or publicize the choice.

The CIDs ask about company organizational charts, business and marketing plans, product roadmaps, bundling of multiple features and products, and pricing, discounting and profitability. Other questions probe the costs and barriers to entering or expanding in markets Microsoft operates in, and solicit documents on the impact of Microsoft's policies or market position. The agency asks some similar questions about AI products, including competition against businesses that combine extra features or services with an AI or software product like Microsoft 365. Kovacic said the references to AI could suggest the FTC considers it a potential part of an illegal tying arrangement or an important competitive advantage.

Outside the US, the European Commission, UK Competition and Markets Authority, and Japan Fair Trade Commission have also been probing the issue, according to The Verge.

AI labs in the dock

In a separate case, four plaintiffs subscribed to ChatGPT, Claude, Grok, or Gemini filed a proposed class-action lawsuit alleging that the developers of these AI models violated antitrust laws when they agreed to slow AI development. Tom's Hardware reported on 21 September that, according to the Associated Press, the lawsuit argues the agreement would "reduce the value consumers get for paid AI subscriptions." It also says coordination started in July 2026, after the leading AI labs signed a statement admitting there is "intense competitive pressure not to unilaterally slow" development.

The plaintiffs recognize the need for AI development to slow for safety, but they say that Anthropic founder Dario Amodei's cooperation proposal is a "shortcut" that "substitutes collective restraint for individual accountability." Attorney Nick Rowley, lead counsel for the plaintiffs, said: "AI will quickly spin out of human control and could kill us all if we allow AI safety and protocol to be controlled by private self-serving agreements between the world's most powerful for profit technology companies." Amodei's essay acknowledged the antitrust risk and indicated he hoped the government would make an exception. OpenAI's Sam Altman responded on X, saying: "We welcome a federal framework that sets consistent safety requirements for frontier AI. But we do not believe we need to wait for an antitrust exemption or legislation to begin the work of providing this confidence." The Trump administration rejected the idea, with the president saying: "AI taking over the World, destroying Humanity, and all other things bad, is a HOAX."

Chinese state media also criticized the announcement, saying the call to slow AI development is a response to Chinese competition, especially as Amodei's essay mentioned slowing China's progress and widening the US gap over Beijing. China Daily called the proposed agreement a "club whose membership rules have been drafted before the guest list is announced" and added that "a global AI-safety framework that excludes China is not quite global."

Three antitrust tracks, three different postures. Google has been found liable and is now under a conduct order it will spend six years complying with. Microsoft is under investigation, with no complaint filed. The AI labs are at the pleading stage, facing a class action whose central claim is that coordination on safety is coordination on price. The common thread is that the remedies in each case are being argued over the same question: whether behavior can be policed without breaking up the structure that makes the behavior profitable.