Brinkema keeps AdX inside Alphabet after ruling Google an illegal monopolist

The choice between structural and behavioral relief is a choice about who pays to notice a violation. A divestiture is expensive once and then enforces itself, because a separately owned exchange has no reason to favour a rival's ad server. Conduct rules are cheap to enter and costly forever, and the cost falls on whoever must detect a breach, which on the open web means publishers and their vendors inferring latency, signal handling and pricing behaviour from outside the system that produces it. Because the court accepted most of the proposed behavioral remedies only "as modified" and did not enumerate them ^[3][6], the firms that will do that detecting cannot yet read which obligations they are meant to police. The board-deck line is that Google kept a small asset and the ad tech complex breathed out. Sizing supports the word small. Wedbush, working from court documents, put Google Ad Manager at 4.1% of Google's overall revenue and 1.5% of its operating profit in 2020, and noted that more recent figures were redacted ^[17]. Ad Manager's share of profit was therefore about 37% of its share of revenue, meaning it earned at roughly a third of the company-average margin ^[26], set against around $294bn of Google advertising revenue last year ^[16]. What the board-deck line omits is that the liability case was never about that profit pool: Brinkema's April 2025 findings turned on wilful conduct in the publisher ad server and ad exchange markets ^[14], where she recorded a durable ad server share above 90% ^[15]. The unresolved question is whether rules can untie what common ownership keeps tied. The operative document does not exist yet, which makes the drafting window the substantive part of this story. The joint filing falls on or about 2 October by the calendar ^[28], and the order fixes no implementation date, names no trustee and sets no duration ^[7]. On duration the two sides arrived seven years apart: Google offered three years of supervision, with later openness to longer, while the government sought ten years of restrictions and a named catalogue of re-monopolisation routes to be policed ^[12][13][29]. There is also a feasibility record the court had in front of it. The Justice Department wanted AdX sold within twelve months of final judgment ^[9], while trial testimony put the open-sourcing of DFP's final auction logic at 24 months on its own ^[11], twice the window the government proposed for the sale ^[27]. A skeptic would say the remedy question is already moot. Much of the ad money has moved to closed platforms including Meta, TikTok, YouTube and Amazon, and Google earns most of its advertising revenue from search rather than from serving ads on websites ^[24]; the consultancy Madison and Wall wrote on Wednesday that equal access, data sharing, the removal of preferential auction rules and monitoring are unlikely to significantly change ad tech's dynamics ^[21]. That reads the aggregate correctly and the unit incorrectly. A publisher still selling open-web display pays Google roughly 20% at AdX ^[10], and the levers actually in play, real-time bidding responses to rival ad servers and the removal of Unified Pricing Rules ^[12], are the ones that move that number. Whether they survived the court's modifications, we do not know yet. This is the third consecutive refusal by a US court to break up a large technology company, after the Meta case over Instagram and WhatsApp and the bid to force a Chrome sale ^[22]. For plaintiffs preparing the Amazon and Apple matters, neither of which is scheduled for trial before 2027 ^[23], the working assumption now has to be that a monopolisation finding buys a conduct decree and a monitor, and that the decree text is where the case is won or lost.