Google Keeps AdX as Antitrust Remedies Land, and the Next Antitrust Fight Is About AI

Judge Leonie Brinkema of the Eastern District of Virginia issued the remedies decision in the ad tech case on 2 September, according to RS Web Solutions, which reported on it on 26 September. Google remains an adjudicated monopolist but keeps the exchange. That distinction matters because the case is now two cases. One concerns what Google must do. The other concerns what Google already did, and that part is heading to appeal.

What the judge actually ordered

Brinkema rejected the DOJ's structural remedies, according to ALM Corp's write-up of the decision. That means no forced sale of AdX, no open-sourcing of the auction logic inside Google's publisher ad server, DFP or DoubleClick for Publishers, and no contingent divestiture of what the industry sometimes calls DFP Remainder. Instead, she adopted most of the behavioural remedies the government proposed, with modifications of her own, and ordered Google and the DOJ to submit a jointly proposed final judgment within 30 days. Her fuller reasoning is expected to stay sealed for about two weeks.

Behavioural remedies change how Google operates inside the auction rather than who owns it. Reporting on the auction conduct points to the removal of "first look" and "last look" provisions, which gave Google's exchange visibility into rival bids before they competed, plus scrutiny of uniform pricing rules that limited how publishers could sell inventory to different buyers. Google had already been phasing some of this out on its own: it eliminated unified pricing in Ad Manager in December 2025, allowing publishers to set buyer-specific floors. The liability finding is untouched. A three-week bench trial in Alexandria in April 2025 concluded that Google had illegally monopolised both the publisher ad server market and the ad exchange market and unlawfully tied the two together, according to Brinkema's ruling at the time. That finding held that Google's conduct had substantially harmed publishers, the competitive process and, by extension, people who consumed content on the open web. Google has said it intends to appeal once the remedies phase concludes.

Both sides claimed something. The DOJ kept its monopoly finding. Google kept AdX. PubMatic, which runs a competing supply-side platform, put out a statement applauding the DOJ's behavioural remedies as a win for competition, ALM Corp reported. The DOJ had argued in a court filing that Google cannot be trusted to honour behavioural commitments; Google's lawyers replied that "distrust is not a reliable basis for abandoning the well-established default of remedial relief in antitrust cases."

A different antitrust argument, about AI

While the ad tech case moves into its remedy phase, a separate antitrust fight is taking shape over AI development itself. Four plaintiffs who subscribed to ChatGPT, Claude, Grok or Gemini filed a proposed class action alleging that the developers of those models violated antitrust law by agreeing to slow AI development, Tom's Hardware reported on 21 September, citing the Associated Press.

The suit argues the arrangement would "reduce the value consumers get for paid AI subscriptions," and says the coordination began in July 2026 after leading AI labs signed a statement acknowledging "intense competitive pressure not to unilaterally slow" development. The plaintiffs accept that slower AI development may be needed for safety reasons, but call Anthropic founder Dario Amodei's cooperation proposal a "shortcut" that "substitutes collective restraint for individual accountability."

"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," said Nick Rowley, lead counsel for the plaintiffs.

Amodei's essay acknowledged the antitrust risk and indicated he hoped the government would grant 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; the president called the notion of AI taking over the world "a HOAX." Chinese state media also criticised the proposal, with China Daily calling it a "club whose membership rules have been drafted before the guest list is announced."

Europe pushes back on the doomer framing

The AI safety panic that produced the cooperation proposal is itself contested in Europe. POLITICO reported on 24 September that European tech executives and cybersecurity professionals are pushing back hard against US-generated warnings about AI extinction, arguing the rhetoric distracts from the practical work of defending systems. James Wise, a partner at Balderton Capital and chair of the UK government's sovereign AI fund, told POLITICO: "Sadly, it has jumped up to the Terminator is coming. I don't think the Terminator is coming at all." Marcus Hutchins, the researcher best known for stopping the WannaCry ransomware attack in 2017, said frontier labs "neither properly contained it by building enough barriers between the agents and the open internet, or adequately separating the agents from each other, nor were they monitoring the systems closely enough to intervene." OpenAI declined to comment and pointed to its report on the incident; Anthropic did not respond to a request for comment.

The timing is awkward for Brussels. Leaders from 22 nations adopted a declaration on Monday saying AI must remain under human direction, oversight and control, and suggested creating a global supervisory regime, POLITICO reported. That sits alongside a G7 meeting and a Trump-Xi summit on Thursday. Europe's cyber community says the Hugging Face incident, which even European Commission President Ursula von der Leyen cited in her State of the EU address, was less a demonstration of runaway AI than a sandbox failure.

There is a hard commercial edge underneath the philosophical argument. Max Junestrand, co-founder and CEO of Legora, an American-Swedish firm building collaborative AI tools for lawyers, called the hacking incidents "a lot of it marketing," but said access to the latest US models is central to his company's own security work. "Security comes even before new functionalities," he told POLITICO.

None of this is happening in isolation. The same week Brinkema's remedy order was circulating, European funding data showed where capital is flowing. Tech.eu's weekly recap, published on 28 September, tracked more than 65 funding deals worth over €2 billion across Europe, with robotics taking €508.3 million, AI €406.6 million and healthtech €305.8 million. Portugal led by country at €510.9 million, ahead of the Netherlands at €435.9 million and the UK at €273.9 million.

Switzerland offers a narrower case study. Tech.eu reported on 24 September that Swiss tech companies raised more than €1.1 billion in the first half of 2026, with the ten largest rounds accounting for about two-thirds of the total. Semiconductors drew around €281 million, energy €223 million and healthtech €220 million. The pattern, concentrated capital in a few capital-intensive sectors, is the one antitrust enforcers on both sides of the Atlantic say they are watching.

On the grid, the US Department of Energy is placing a different kind of bet. Canary Media reported on 24 September that the agency awarded $1.9 billion under its SPARK grant programme to 31 projects across 26 states for advanced transmission technologies, with participating utilities pledging $3.35 billion in matching funds for a total of $5.25 billion. Recipients include American Electric Power, CenterPoint Energy, Duke Energy, Eversource Energy, Rocky Mountain Power and PPL Electric Utilities. Energy Secretary Chris Wright said in a Thursday statement that the grants will "get more out of the infrastructure we already have, move more electricity across the grid, and help deliver affordable, reliable, and secure power that will fuel American prosperity for decades to come."

What to watch

• The joint proposed final judgment from Google and the DOJ, due within 30 days of the 2 September order.
• Brinkema's unsealed reasoning, expected about two weeks after the order.
• Google's appeal of the liability finding once remedies conclude.
• Whether the AI cooperation lawsuit survives early motions, and how the labs respond to the "self-serving" characterisation.
• Whether Europe's regulatory push, including the 22-nation declaration, translates into enforcement that touches US AI vendors.

The through-line is that antitrust has stopped being a single-industry story. Ad tech produced a remedy that changed conduct but not ownership. AI is producing a liability theory aimed at the terms on which models get built. And the money keeps moving: over €2 billion in a single European week, according to Tech.eu, into exactly the capital-heavy sectors regulators say deserve scrutiny.