The antitrust case has delivered its final verdict, why wasn't Google's advertising business broken up?

I

On September 2, 2026 local time, Judge Leonie Brinkema of the United States District Court for the Eastern District of Virginia ruled that Google is not required to sell its ad exchange platform AdX, nor to divest its ad server DFP.

The two products together form Google Ad Manager, commonly referred to in the industry as GAM.

For those unfamiliar with programmatic advertising, this system can be understood as a large farmers' market.

DFP is like the stall management system in the market, helping websites and media manage ad slots, determining where stalls can be set up, when they are for sale, and the minimum price for them. AdX, on the other hand, is like a trading hall that brings together advertisers who want to buy ads and media who want to sell ad slots; the role of GAM is to integrate stall management and the trading hall into a single system.

Here, Google not only runs the market, but also controls a large number of buyers, and can view all the quotes and transaction results in the market.

It acts both as an athlete and a referee, and also participates in formulating the rules of the game.

In April 2025, the U.S. federal court had already determined that Google had formed an illegal monopoly in the open web ad technology market by bundling ad servers and ad exchanges, prioritizing its own products, and restricting competitors.

This ruling discusses how to deal with the situation after the violation is confirmed.

The U.S. Department of Justice hopes the court will carry out a thorough "surgical operation": requiring Google to sell AdX, and further divest DFP if necessary, to structurally eliminate the conflict of interest arising from Google's simultaneous control of multiple links in ad transactions.

According to the final result, Judge Brinkema did not adopt this plan.

She opted for behavioral remedies, which means not splitting the company or selling assets, but imposing new rules on Google's business operations.

The full ruling has not been made public yet. Brinkema's written opinion will be sealed for 14 days to allow both parties to the lawsuit to check for information that should not be disclosed. The outside world does not yet know the specific content of each rectification measure, only that the court has adopted "most" of the behavioral remedy proposals put forward by the Department of Justice and Google.

These measures roughly cover three directions:

First, restrict Google from favoring its own products in ad auctions;

Second, require Google to open more auction data to media;

Third, require Google to treat other ad exchange platforms and ad servers equally.

Google previously voluntarily proposed to allow competitors to view part of the auction information in AdX in real time, and allow media to access independent auction tools between DFP and AdX. In this way, Google's own system will no longer have the final say in every ad transaction.

Google also proposed to cancel the unified pricing rule, allow media to set different minimum prices for different bidders, and promise not to use mechanisms such as "priority view" and "last view" to adjust auction results.

In short, in the past Google might first look at other people's bids before deciding how much its own platform would offer. What the new rules aim to do is to make Google and other platforms play their cards at the same time as much as possible, preventing it from acting after seeing others' hole cards.

However, the opening measures proposed by Google mainly target ordinary display ads on websites, and do not include video ads, streaming TV ads and in-app ads.

The Department of Justice believes that the scope of opening should be larger. Advertisers should also have access to corresponding auction information, Google should not adjust the auction path or price in any way to favor its own products, and the relevant program code should also be subject to external inspections. Independent auction tools should not only be used for the types of ads that Google is willing to open up.

Which issues the court ultimately sides with Google on and which it sides with the Department of Justice on will determine whether this ruling is just a minor tweak or can change the way the ad market operates.

II

Brinkema's judgment had already shown a tendency during the hearing at the remedy stage.

What she values more is who will bear the cost after the split, who can take over the business, and how long this "operation" will take.

The first issue is that small and medium-sized media may be accidentally injured first.

At present, a large number of websites and small publishers use DFP for free. According to data provided by Google's lawyers in court, 92% of publishers use this ad server for free.

For large media groups, replacing ad servers is troublesome but not impossible. They have their own technical teams and the ability to connect with multiple ad platforms at the same time.

Small websites do not have such conditions. Many long-tail media rely on free DFP to manage ad slots, and then find ad buyers through AdX. If the two products are forcibly separated, the new company starts charging, the interface changes, or the service level drops, the first to be impacted will be these small media with no bargaining power.

This creates an awkward situation: the split, originally intended to protect market participants, may first harm the weakest group of people in the short term.

The second issue is who will take over the business.

The court can order Google to sell AdX, but cannot create an ideal buyer out of thin air.

Microsoft seems to have the capital, technology and ad business foundation, but if Microsoft acquires AdX, the transaction itself is likely to trigger antitrust review again. The market may end up just shifting from "Google controlling the key ad platform" to "Microsoft controlling the key ad platform".

Moreover, AdX is not an office that can be moved away directly. It is deeply connected with Google's ad servers, advertiser products, data systems and engineering teams. During the split, it is necessary to divide the code, data, personnel, customer contracts and technical interfaces, and determine how both parties will continue to provide services after the split.

On the surface, it is selling a business, but in reality it is like dismantling an engine from a flying plane and replacing it with a new operation team.

The third issue is time.

Mandatory divestment is the most severe antitrust penalty, and Google will almost certainly appeal. Once the appeal procedure is entered, the split order is likely to be suspended. The two sides will continue to argue about which assets to split, who to sell to, how to isolate data, and how to ensure that the business is not interrupted.

The whole process may last three to five years, or even longer.

However, the ad technology market is developing too fast under the influence of AI.

The importance of traditional web display ads is declining, while video, streaming TV and in-app ads are growing continuously.

At the same time, generative AI search is reducing the number of users clicking on media websites, and media traffic and ad revenue are also under pressure.

By the time the split is completed, the market that the court is trying to solve today may have evolved to version 2.0 or 3.0, or even be completely unrecognizable.

In contrast, behavioral remedies can take effect faster. Even if Google continues to appeal, some rules may begin to be implemented within one to one and a half years.

The Department of Justice, as the plaintiff, obviously does not agree with this.

Lawyers from the Department of Justice believe that splitting is a cleaner, lower-risk and once-and-for-all solution. As long as Google continues to control both the ad server and the ad exchange, the conflict of interest will not disappear. Behavioral remedies only tell Google what not to do, but do not take away its ability to do those things.

More importantly, the Department of Justice believes that Google is not trustworthy.

Programmatic ad auctions are often completed within hundreds of milliseconds, forming a huge black box, and it is difficult for the outside world to see the order of bids, data calls and traffic allocation in each transaction.

Even if Google ostensibly complies with the rules, it may continue to give its own products advantages through algorithms, interfaces, delays or default settings.

Regulators cannot sit beside every auction to supervise.

Google's lawyers responded that distrust of a company cannot be a reason to bypass established antitrust principles. Structural split must target the violations already determined by the court, and prove that milder remedial measures cannot solve the problem.

Google also emphasized that the monopoly determined by the court mainly occurs in the open web display ad market, but AdX and DFP also process video and in-app ads, and Google has not been found to have an illegal monopoly in these markets. If the overall divestment is carried out, it is equivalent to penalizing the businesses that have not been convicted of violations together.

In the end, Judge Brinkema accepted Google's position on the issue of "whether to split", while retaining measures such as opening up data, restricting self-preference, and accepting continuous supervision.

There is a clear divergence within the industry on this.

PubMatic, an independent ad platform competing with AdX, has a relatively pragmatic attitude. It believes that as long as the court can strictly implement behavioral remedies, there is an opportunity to allow different ad platforms to compete on more equitable terms. Compared with a split lawsuit that lasts for many years, changing the auction rules immediately may be more practical.

Another group of industry insiders are more pessimistic.

Jay Friedman, co-founder of CartographAI and former CEO of the Goodway Group, raised a question: If a media does not want to use Google's ad server, but still wants to access the ad demand controlled by Google, does it really have a choice?

If the answer is no, the so-called open competition is still built on Google's system.

Shamsul Chowhury, Senior Vice President of Paid Media at Zeno Group, used a more vivid metaphor: Google owning both the ad serving platform and the performance measurement platform at the same time is like letting students grade their own homework.

Meta, Microsoft, Amazon, The Trade Desk and other independent ad technology companies certainly hope to see Google split. In addition to fair competition, they can also take more digital ad budgets away from Google.

However, these companies themselves are also facing antitrust investigations, so the court will not easily believe that replacing one giant to take over the business can solve the problem.

Split seems thorough, but it has high cost, long time and unpredictable results; behavioral rectification takes effect faster, but relies on long-term supervision, and regulators may not be able to keep up with Google's technological changes.

The court chose the latter, leaving the most difficult problem to the implementation stage.

III

In the short term, this ruling first brings stability.

Media do not need to replace ad servers immediately, and advertisers and agencies do not need to rebuild their ad delivery links. For an industry that processes massive real-time bids every day, where any system interruption may cause revenue loss, avoiding forced split reduces the risk of technology migration.

But the other side of the coin is that Google's core position has not changed.

GAM, AdX and the advertiser demand controlled by Google still remain in the same system. Even if media do not like Google, it is difficult for them to completely leave Google. Especially for small and medium-sized websites, GAM, which is free, stable and has a large number of buyers, is still the most realistic choice.

The potential benefit for media is more auction information and pricing power.

If the court requires AdX to open bids in real time and allow media to set different reserve prices for different bidders, media can clearly see how much each party offers, and allow multiple ad platforms to compete more fully.

In the past, media often only saw the final transaction result, but did not know how much information advantage Google had in the auction process. If data is more transparent in the future, media may have the opportunity to increase ad revenue, and can more accurately judge the real performance of different platforms.

Independent SSPs and ad exchange platforms may also get new opportunities.

If AdX must comply with the same rules as competitors such as PubMatic and Magnite, Google's internal advantage formed between the ad server and the exchange platform will be weakened, making it less difficult for independent platforms to access media ad systems.

Open auction tools such as Prebid may also become beneficiaries.

Prebid can be understood as an independent auction table that allows multiple ad platforms to bid at the same time. If the court requires Google to connect to this table more deeply, AdX will no longer be able to bid according to its own rules in its own room forever.

But this does not mean that independent platforms can immediately stand on an equal footing with Google.

Google not only controls the exchange platform, but also has huge advertiser demand, search and video traffic, user data and ad performance measurement tools. Even if the auction rules become more fair, it is difficult for other platforms to make up for these resources quickly.

For advertisers and agencies, the biggest potential change is increased transparency.

Advertisers may more easily know how much of their 100-yuan ad fee goes to the media, and how much is taken away by the exchange platform and intermediate service providers. They can also more clearly judge whether Google recommends its own products because of better performance, or because Google controls the rules.

However, the new openness requirements will also increase system transformation costs. Google and other ad platforms need to adjust interfaces, auction logic and data reports, while media and agencies need to retest transaction rates, revenue changes and traffic quality.

In the short term, the industry may enter a period of rule adaptation.

A more far-reaching impact is that future ad technology regulation may no longer take split as the main goal. This means that future regulation will shift to requiring large platforms to open interfaces, share data and prohibit self-preference.

The focus of competition in the programmatic ad industry may also change accordingly. Platforms cannot only rely on closed ecosystems and exclusive resources to retain customers, but also need to compete more on algorithm efficiency, data quality, service capabilities and supply chain transparency.

But all this has a premise: the court must be able to enforce the rules.

Programmatic ad systems are too complex. Google may not openly reject competitors, but it can make the other party's interface a little slower, the data a little less, and the default position a little worse. Each adjustment seems to be an ordinary product design, but cumulatively it can determine who wins the auction.

This article is from the WeChat official account "Daoke Doc", author: Daoke Doc, published with authorization from 36Kr.