How Google’s Antitrust ‘Win’ May Hobble Its AI Future

• Google avoided a breakup of Chrome and Android, but new court-ordered limits on exclusivity, data and ad access weaken its AI moat.
• The ruling lowers major barriers for rivals such as OpenAI and Perplexity by expanding distribution, ad monetization, and limited data access.
• Generative AI reshaped the case, with remedies aimed at preventing Google’s search dominance from carrying over into the AI era.

When U.S. District Court Judge Amit Mehta ruled on Sept. 2 that Google did not have to divest its Chrome web browser or Android operating system, Wall Street celebrated what it deemed to be the company dodging the worst-case scenario – and boosted the stock up 9% the following day.

But the ruling on the U.S. government’s antitrust case against Google was a win for legacy Google. For the Google that wants to lead in an AI future, the answer is still up for grabs.

In 2020, the U.S. government sued Google on antitrust grounds, alleging that it acted as a ‘monopolist’ by paying billions of dollars to browser developers, device manufacturers and wireless carriers to be their default general search engine. Altogether, they were where people used search the most: phones, laptops and web browsers. By doing so, Google locked out the competition. Wall Street feared a break-up of Google – like AT&T in 1984 – by requiring it to divest Chrome and Android.

In his ruling, Mehta preserved Google’s basic corporate structure but ordered changes to pry open three areas that have long kept competitors out: exclusive deals, search data and ads shown alongside search results. The ruling limits exclusive deals that made Google the default search engine on the world’s devices, forces the company to give others access to both search results and search text ads, and mandates limited data sharing that can help competitors train and evaluate AI systems.

Together, those measures weaken Google’s AI moat without detonating its legacy business. Hence, the irony: The rise of generative AI helped Google dodge a breakup, but the court’s remedies are designed to make its AI competitors more viable going forward.

A BofA Global Research report said Google won in three areas: it was able to keep Chrome and Android; the revenue-sharing model is largely intact, preserving a core distribution strategy; and it will not be forced to share query-level advertising data with rivals or advertisers.

Together, those measures weaken Google’s AI moat without detonating its legacy business.

The report cited the following negatives: Data-sharing mandates will further increase search competition; loss of exclusivity among device partners could weaken Google’s competitive moat; syndication (giving rivals access to its data and ad supply) will increase competition and may impact traffic growth; and an uncertain impact on Gemini and future gen AI integrations.

In his 230-page opinion, Mehta framed the legal sanctions in light of advances in AI. “The emergence of GenAI changed the course of this case,” he wrote. The ruling “has been as much about promoting competition among GSEs (general search engines) as ensuring that Google’s dominance in search does not carry over into the GenAI space.”

The court barred exclusive contracts covering Google Search, Chrome, Google Assistant and the Gemini app, and limited revenue-sharing deals with Android device makers and wireless carriers to one year.

Moreover, the court said that a device maker or wireless carrier may “simultaneously preinstall Google Search and a non-Google GenAI product, like ChatGPT, Perplexity, or Claude,” or a rival general search engine and the Gemini app. That change matters because now there are more ways to search, such as Google’s ‘Circle to Search’ and Google Lens.

Ad monetization was the second moat the court weakened. Google makes money from ads that are shown alongside search results. But it controls access to the ad marketplace. The court’s decision means search rivals can more easily plug into the ad marketplace to make money.

Data was the third moat affected. Google crawls the world’s web pages so it has the largest and most comprehensive dataset – curated, ranked, deduplicated and constantly updated – for its search results. Search rivals have crawled parts of the internet while AI startups have used public web scrapes to train their LLMs, but that’s not the same as building a high-quality, continuously updated search index.

What it means for rivals: The ruling lowers three big barriers to entry at once.

The judge is forcing Google to let rivals such as ChatGPT or Perplexity do a one-time snapshot of specified elements of Google’s search index, plus limited user-interaction data. However, they won’t know Google’s ranking system or get its LLMs that interpret queries and decide what’s most relevant.

Google was concerned that rivals would be able to mimic its search rankings, which are the most used in the world, according to the ruling. So the judge settled on one-time “snapshot” plus limited user-interaction datasets that can help rivals, but without cloning Google. “Helping competitors improve their search engines is precisely what this remedy is designed to accomplish,” the judge wrote.

The court did leave Google room to keep paying partners – the judge warned that a blanket payment ban could impose “crippling” downstream harms – but the one-year limit and anti-tying language turn those checks into annual, nonexclusive renewals instead of multi-year exclusivity deals.

What this means for Google: The company keeps its structure and can still negotiate distribution, but its default strategy now refreshes annually and must coexist with other engines and AI assistants.

What it means for rivals: The ruling lowers three big barriers to entry at once.

First, distribution: Device makers and carriers can place non-Google AI assistants next to Google apps and still take Google’s payments, at least for a one-year term. That gives players like OpenAI, Anthropic and Perplexity a path to ride new access points – side buttons, wake words, Lens-style camera search – without triggering contract tripwires.

Second, monetization: Ad and search syndication can help AI-first products bridge to a sustainable business while they scale their own models and ad tech. Third, data: The one-time index snapshot and limited user-interaction datasets won’t replicate Google’s quality overnight, but they can speed long-tail coverage and evaluation, especially for retrieval-augmented generation.

The judge said the order runs for six years, as generative AI is “breaking barriers seemingly at light speed.”