A judge refuses to break up Google ad tech and instead bans first look, last look and Unified Pricing Rules for six years

The memorandum opinion explaining why Google will not have to sell its advertising exchange came out of seal on 16 September 2026. It runs to 106 pages, and neither party moved to redact any of it. Judge Leonie M. Brinkema had entered the remedies order itself two weeks earlier, on 2 September, in United States of America et al. v. Google LLC, case 1:23-cv-108 (LMB/JFA) in the Eastern District of Virginia, docket entry 1857. For those two weeks the public could read that the government had lost its bid to break up Google's ad tech business, and could not read why.

The court found structural relief, in its own words, "neither realistic nor needed".

What was refused

Three structural remedies were rejected outright: divestiture of the AdX exchange, open-sourcing of the final auction logic inside DoubleClick for Publishers, and a contingent divestiture of the remainder of DFP had the behavioural measures failed.

The reported reasoning is practical rather than doctrinal. According to trade accounts of the opinion, the court weighed that no obvious buyer for AdX exists, and that the most capable acquirer would attract antitrust scrutiny of its own; that a divestiture running three to five years would be overtaken by a market moving faster than that; that appeals would delay any relief for years; and that small publishers who currently use DFP at no charge could be harmed by breaking the service apart. Behavioural remedies, by contrast, are described as enforceable within a year and fully implementable in twelve to eighteen months. [UNVERIFIED: the 106-page opinion has not been read in the original for this piece. Every characterisation of the court's reasoning here comes from AdExchanger's and ppc.land's accounts of it, not from the document.]

There is something worth noticing in that list. Two of the four stated objections are consequences of the conduct the court already found unlawful in April 2025. AdX is hard to separate because it was built to be inseparable, and small publishers are exposed because free DFP is how the tie was maintained. Entanglement became the argument against undoing the entanglement.

What was ordered

The behavioural package is narrower than a breakup and considerably more specific than the phrase "behavioural remedy" usually implies.

Google must stop three auction practices by name. "First look" gave AdX the first opportunity to bid on a publisher's impression before rival exchanges could see it. "Last look" let AdX see the highest competing bid before submitting its own. Unified Pricing Rules, introduced in 2019, stopped publishers setting a higher minimum price for AdX than for its competitors. Google must also stop withholding real-time bid data from rival ad servers.

Google must build API integrations connecting AdX and DFP to Prebid, the open-source header bidding framework, so that Prebid can solicit real-time AdX bids across indirect open-web display inventory. The integrations have to be "functionally equivalent", which is the court closing the obvious escape route of shipping a deliberately worse connector. AdX must submit real-time bids into competing publisher ad servers on the same terms it offers DFP. Google must publish technical documentation explaining how DFP selects a winning bid, including pricing and bid adjustments, and share win and loss bid data with publishers.

On the buy side, AdWords may not bid directly into DFP, may not favour Google's own ad tech, and is restricted in using first-party data to advantage Google's infrastructure. DV360 was left alone: the court found an insufficient connection to the conduct at issue, notwithstanding that the opinion discusses Project Poirot, the 2017 programme reported to have cut Google's bids into rival exchanges through DV360 while bidding at full price through AdX.

The obligations run for six years, a term Google itself proposed and the court adopted. An independent technical monitor is appointed for the same six years, with access to Google's employees, systems and source code; the Justice Department had asked for fifteen. The remedies apply globally rather than only in the United States, and take effect sixty days after judgment. A jointly proposed Final Judgment is due on 2 October 2026. Google is expected to appeal on entry, and is separately appealing the underlying liability finding of 17 April 2025.

Why it matters

This was the clearest test in a decade of whether a court would structurally decentralise a piece of privately owned critical infrastructure, and the answer was no. The intermediary stays intact, keeps both sides of the market, and is instead placed under supervision. That is a different theory of remedy, and it is worth being precise about the difference rather than treating it as a defeat with extra steps.

Breaking a chokepoint into parts is a one-time act that then requires no further state involvement: the parts either compete or they do not. Forcing an open interface onto a chokepoint leaves the chokepoint in place and requires someone to keep watching it. Here that someone is a court-appointed monitor with source code access for six years, which is a genuinely unusual degree of continuing state oversight of a private system, and which expires on a date certain while the market position it polices does not.

The Prebid mandate is the interesting part, because it is not conduct regulation in the ordinary sense. Prebid is the open-source alternative the publishing industry built precisely to route around Google's auction, and Google spent years declining to participate in it. The order compels the incumbent to plug into the commons its competitors assembled. If interoperability mandates can substitute for separation, this is the case that will show it, and there is a six-year clock and a public docket against which to check.

The sealing sequence is a smaller point that should not be skipped. A ruling on the ownership and governance of infrastructure that intermediates a large share of the open web was public as an outcome for fourteen days before it was public as an argument. It was eventually released whole, with no redactions sought by either side, which raises the question of what the seal was protecting in the first place.

What is still unknown

Whether the Prebid integration requirement survives appeal, and whether the six-year term begins to run during an appeal or after it.

How the technical monitor is selected, who pays, what the monitor may publish, and what happens at the end of six years if the market position is unchanged.

[NEEDS DATA: the text of the jointly proposed Final Judgment due 2 October 2026, which converts the opinion into operative obligations.]

Whether the European Commission's separate ad tech proceeding against Alphabet, which has pointed toward divestiture, now diverges from the American outcome, and what a company subject to both does.

[NEEDS DATA: any measure of AdX's share of open-web display before and after the sixty-day effective date, which is the only way to judge whether the interoperability remedy worked.]

Sources

ppc.land on the 2 September order, the case number and docket entry, the rejection of all three structural remedies and the sealed opinion: https://ppc.land/doj-loses-adx-divestiture-bid-as-brinkema-accepts-behavioral-remedies/

ppc.land on the unsealed opinion, the Prebid integration requirement, the bid data obligations, the AdWords restrictions, the six-year monitor and the global scope: https://ppc.land/unsealed-ruling-orders-google-to-open-adx-and-dfp-to-prebid/

ppc.land on the court's stated objections to divestiture, the practices Google must stop, the six-year term and the twelve to eighteen month implementation window: https://ppc.land/judge-spares-googles-ad-exchange-and-rewrites-its-auction-rules-instead/

AdExchanger's summary of the unsealed 106-page decision, the absence of redactions, the DV360 finding and the fifteen-year monitor the Justice Department had sought: https://www.adexchanger.com/platforms/the-court-just-unsealed-judge-brinkemas-remedies-decision-in-the-google-ad-tech-antitrust-case-heres-your-tldr/

AdExchanger's earlier report of the 2 September ruling: https://www.adexchanger.com/antitrust/google-wont-have-to-break-up-its-ad-tech-business-judge-brinkema-rules/