- A bipartisan amicus brief filed August 4, 2026, argues Google’s search monopoly revenues fund its AI research and distribution deals at a scale most competitors cannot match.
- The brief cites Apple’s January 2026 agreement to use Google’s Gemini models and cloud infrastructure as evidence that search-era distribution leverage is already replicating in AI.
- Both the Vanderbilt Policy Accelerator brief and a parallel Open Markets Institute filing contend the proposed antitrust remedies are too narrow to address Google’s emerging AI distribution advantages.
A bipartisan amicus brief filed August 4, 2026, in case No. 26-5023 makes an argument the DOJ’s proposed remedies have not fully confronted: Google’s search monopoly is not just a competition problem in itself, it is the financial engine behind Google’s AI ambitions. Without stronger intervention, the filing contends, the distribution leverage that secured Google’s search dominance will play out again in generative AI.
Search Revenue as the AI War Chest
The brief, advanced by Joel L. Thayer and Asad Ramzanali of the Vanderbilt Policy Accelerator, makes a structural argument: generative AI cannot credibly challenge Google‘s search monopoly because that monopoly is what finances Google’s AI ambitions. Search revenues fund AI research, infrastructure and distribution deals at a scale most competitors cannot match, creating a self-reinforcing dynamic where dominance in one market directly subsidises a bid for dominance in the next.
Why Current Remedies Fall Short
The brief joins a line of criticism targeting the original antitrust case’s proposed remedies as too narrow. The Open Markets Institute filed a parallel brief on the same date making related arguments. Both filings argue that allowing Google to continue paying billions for default search placement, with Apple, browser developers and device manufacturers, preserves the distribution infrastructure the court already deemed unlawful.
Those payments do more than secure search placement. According to the brief, they cement long-term partnerships that then shape which AI models and cloud services those partners adopt. The Apple deal is the clearest example. On January 12, 2026, Apple agreed to use Google’s Gemini models and cloud infrastructure to power AI features including a Gemini-assisted Siri, according to reports. Thayer and Ramzanali cite this as evidence that search-era distribution leverage is already replicating itself in AI, a concern the EU AI Act’s enforcement framework addresses from a different regulatory angle.
The brief’s concern is not where Google stands today but where those distribution deals could take it.
The 21% Figure
The brief references a figure tracing to Polymarket prediction-market data reported by PPC Land, where bettors currently give Google a 21% implied probability of having the best AI model, down sharply from 71% in February. That drop coincided with Anthropic overtaking Google in early March, and Anthropic now leads the market at 65% implied probability. It’s worth being precise about what this number is: a sentiment measure reflecting betting activity on which company will have the top model, not a usage or market-share statistic. The brief’s own framing, treating the figure as a potential or projected position rather than a current market reading, is consistent with that distinction.
Google’s Position
In previous filings, including its legal defence during the DOJ antitrust case, Google has characterised structural remedies such as divesting Chrome or Android as harmful to consumers and anticompetitive in their own right.
The outcome of the case will help determine whether existing antitrust law can reach forward-looking market dynamics: not just the search monopoly the court has already found, but the AI distribution advantages it may already be generating. That question puts this proceeding in the same territory as binding DMA orders the EU has issued against Google’s Android and Search a parallel track with its own enforcement teeth. Our AI Policy coverage follows how these rules land in practice.
