EU Google Crackdown Quietly Rewards US AI R

EU's Google Search Crackdown Quietly Rewards American AI Rivals

The European Commission's Digital Markets Act ruling, due by July 27, 2026, will force Google to share search data with rivals on FRAND terms — but the companies best positioned to exploit it are not European search engines but American AI competitors like OpenAI, Perplexity, and Microsoft, who need Google's query and click data to power the AI search tools eroding Google's core business.

The European Commission is expected to announce as early as next week that Google violated the bloc's Digital Markets Act by steering users toward its own shopping, travel, and hotel services in search results, according to the Financial Times. The decision, targeting Alphabet under Article 6(5) of the DMA, carries fines worth hundreds of millions of euros across two cases, with the threat of daily penalties if Google fails to comply within 60 days. But the more consequential half of the crackdown is a separate specification proceeding — case DMA.100209 — that will force Google to hand anonymized search data to rival search engines and AI chatbots on fair, reasonable, and non-discriminatory terms, with a final binding decision targeted by July 27, 2026.

From display boxes to data taps

The self-preferencing case has been building since March 19, 2025, when the Commission released preliminary findings that Google Search promotes Alphabet's vertical services — Google Flights, Hotels, and Shopping — over competitors like Kayak and Skyscanner, while Google Play prevents app developers from directing users to better offers outside the platform, according to the Commission's case documentation. Alphabet submitted a compliance report on March 7, 2025, claiming it had introduced "enhanced individual free web result formats for comparison services and merchants" on an opt-in basis. The Commission was not persuaded. It opened formal proceedings examining whether those measures "can undermine effective compliance" with Article 6(5) because they "do not fully address possible favourable treatment in ranking of Google's services on the SERP."

The data-sharing remedy is the structural escalation. On January 27, 2026, the Commission opened specification proceedings under Article 6(11) of the DMA, which requires gatekeepers running search engines to provide third parties "access on fair, reasonable and non-discriminatory terms to ranking, query, click and view data in relation to free and paid search," with personal data anonymized, as set out in the consolidated DMA text. The Commission's preliminary findings, adopted April 16, 2026, specify that Alphabet must share all query, view, click, and ranking data it collects to optimize its own services — at parity with what it uses internally — via an API or equivalent tool, for at least five years per beneficiary, as detailed in the Commission's preliminary measures.

The anonymisation requirements are granular: no user account information, no search histories, no precise timestamps, very long and rare-word queries suppressed, location generalized, precise interaction durations replaced by time intervals. Data must be shared with a latency of at least seven days, and FRAND pricing corresponds to incremental cost-based pricing — eligible beneficiaries pay only Alphabet's costs plus a reasonable return on capital, with micro and small enterprises exempted from paying more than incremental costs, per the Commission's case summary. The pricing measures will apply for five years, after which Alphabet may renegotiate — but any resulting terms must remain FRAND. For beneficiaries that are themselves designated gatekeepers, Alphabet may deviate from the standard incremental cost methodology and negotiate separately, provided terms remain fair and non-discriminatory.

The scope is deliberately broad. The Commission's preliminary findings on eligibility specify that beneficiaries must operate an online search engine in the EEA, "including AI chatbots that provide online search engine functionalities." That phrase — added in the April 2026 measures — is the hinge on which the entire strategic significance of the case turns. It means the data tap is not designed for Ecosia or Qwant alone. It is designed for any AI-powered search service with the technical capacity to ingest query, click, and ranking data at scale.

The 2017 precedent that haunts this case

The Commission has been here before, and it lost. In 2017, Brussels fined Google €2.42 billion for self-preferencing its Google Shopping comparison service and ordered the company to treat rival comparison shopping services no less favorably than its own, under Article 7(1) of Regulation 1/2003. Google complied by introducing an auction-based mechanism where rival comparison services bid for placement in shopping boxes atop search results.

It did not work. A study of 25 comparison shopping services across 21 European markets — including market leaders in seven of the 13 commercially most relevant countries — found that less than one percent of traffic was directed to competing shopping services three years after the remedy, as documented in academic analysis published in the Journal of Antitrust Enforcement (Taylor & Francis). The study, based on empirical data covering consumer traffic, conversions, revenues, and customer reactions, concluded that Google's chosen compliance mechanism did not reflect equal treatment of rival comparison shopping services. The European Consumer Organisation, BEUC, argued that competition had not been restored because Google's algorithms continued to downgrade rivals through the addition of new ranking criteria, and called for a structural remedy separating Google's search engine from its comparison shopping services.

By 2019, then-Competition Commissioner Margrethe Vestager publicly acknowledged that the Commission did not see much traffic flowing to rival comparison services. The Court of Justice upheld the €2.4 billion fine in September 2024, but the behavioral remedy's failure is now the intellectual foundation for the DMA's ex-ante approach. Where the 2017 case relied on Article 102 TFEU (abuse of dominance) and post-hoc enforcement, the DMA imposes proactive obligations on designated gatekeepers — fines of 4 to 20 percent of global annual turnover for non-compliance, as CSIS analysis notes. Seven gatekeepers have been designated — Meta, Alphabet, Amazon, ByteDance, Apple, Booking, and Microsoft — covering 23 core platform services.

The Commission's own Staff Working Document on the DMA review, published April 28, 2026, acknowledges the enforcement gap. The document notes that "many vertical search service providers and trade associations have emphasised that Article 6(5) is not being enforced with sufficient rigour" and that "enforcement must be accelerated and strengthened to prevent irreversible" harm, according to the Commission's DMA review. The data-sharing remedy is the structural answer the 2017 case lacked. Instead of telling Google how to display rivals, the Commission is forcing Google to hand over the raw material — query, click, ranking, and view data — that any rival needs to compete on search quality at all.

Who actually benefits: the American AI angle

The Commission frames the data-sharing remedy as a tool to help European search rivals compete. But the companies best positioned to exploit a firehose of Google's anonymized search data are not European startups. They are American AI firms — OpenAI, Perplexity, Microsoft's Bing-powered Copilot — that are building AI search tools to compete directly with Google's core business. These companies have the engineering capacity to ingest large-scale query and click data and use it to train and fine-tune retrieval-augmented generation systems, the architecture underpinning AI search. A small European search engine like Qwant, with a fraction of the resources, cannot match that data-processing capacity even with free access to the data tap.

This is the non-obvious consequence: a remedy designed to curb an American gatekeeper may primarily strengthen other American incumbents who are already eroding Google's search dominance from the AI side. The core debate inside Brussels is precisely whether data sharing levels the playing field for European startups or mainly benefits US AI giants that rely on Google's index, as Politico reported. The Commission's pricing measures try to address this asymmetry: micro and small enterprises are exempted from paying more than incremental costs, while beneficiaries that are themselves designated gatekeepers — Microsoft, if Bing were designated — could face higher negotiated rates. But the structural advantage of scale remains. The company with the most advanced AI infrastructure wins the most from raw data access, regardless of where it is headquartered.

The Commission's April 2026 press release frames the measures as promoting "contestability and non-discrimination" in search, not as industrial policy for European champions, according to the Commission's announcement. That framing is deliberate. The DMA is competition law, not trade policy. But the effect is to subsidize the data needs of the very companies — OpenAI, Perplexity, Microsoft — that the Commission's own AI Act and industrial policy documents identify as competitive threats to European digital sovereignty.

The parallel AI probe

The data-sharing case does not exist in isolation. On December 9, 2025, the Commission opened a separate antitrust investigation into whether Google's AI Overviews and AI Mode features, and its use of YouTube content for AI training, amount to abuse of dominance or self-preferencing, with potential penalties up to 10 percent of global turnover, Reuters reported. The probe examines whether publishers are compelled to allow use of their content for AI without adequate compensation or a meaningful opt-out, and whether rivals face access restrictions to the same content — a form of data foreclosure that mirrors the search data case, per EU Digital Law analysis.

Read together, the two actions reveal a coherent strategy: the Commission is attacking Google's data advantage from two directions. The DMA data-sharing case forces Google to give rivals access to the output of its search engine — what users click and query. The AI antitrust probe attacks Google's input advantage — the content it crawls and uses to train AI models. Both aim to ensure that Google cannot leverage its search dominance into the next generation of AI-powered discovery. The Commission is separately weighing whether Google should open the same features used by its Gemini assistant to third-party AI providers, a signal that the confrontation is moving beyond traditional search into AI-mediated discovery, as Noah Intelligence reported.

This dual-track approach also addresses a gap in the 2017 Shopping precedent. That case attacked self-preferencing in one vertical — shopping comparison — and lost. The DMA cases attack the structural data moat across all verticals simultaneously, and add an AI layer that did not exist in 2017. The Commission is not repeating its mistake. It is widening the battlefield.

The ad tech escalation

The Google Search cases follow a pattern of escalating penalties. In September 2025, the Commission fined Google €2.95 billion for abusing its dominance in ad tech by favoring its own advertising exchange, AdX, over competing exchanges where ads are bought and sold in real-time, as the BBC reported. Competition Commissioner Teresa Ribera said the fine was increased because "this is the third time Google breaks the rules of the game" and warned that the company had 60 days to detail how it would change its practices or the Commission would impose its own solution. Ribera added: "At this stage, it appears the only way for Google to end its conflict of interest effectively is with a structural remedy, such as selling some part of its ad tech business."

That was the first time a senior Commission official publicly floated a breakup remedy. The DMA search cases do not explicitly threaten structural separation, but the Commission's Staff Working Document notes that "Article 16 of the DMA maintains the status quo where behavioural remedies are preferred over structural ones and only the failure of behavioural remedies may allow for a structural remedy." The data-sharing specification is the Commission's attempt to make a behavioral remedy work. If it fails — as the 2017 Shopping remedy failed — the political case for structural separation strengthens.

The cumulative financial pressure is mounting. Google has now paid or been ordered to pay over €9.4 billion in EU antitrust fines: €2.4 billion for Shopping (2017), €4.1 billion for Android (2018, upheld July 2, 2026, per The Economist), and €2.95 billion for ad tech (2025). The DMA fines — if imposed — would add to that total, with the possibility of daily penalties for ongoing non-compliance.

The UK converges

The EU is not acting alone. On June 17, 2026, the UK's Competition and Markets Authority ordered Google to provide greater transparency on how search rankings work, to rank results using "objective and non-discriminatory criteria," to introduce clearer complaint processes, and to allow users to transfer search data to authorized third parties, as Al Jazeera reported. Google accounts for more than 90 percent of UK search queries. The CMA also granted UK publishers the right to opt out of Google's AI search results, a world-first requirement giving news organizations leverage to negotiate content deals, per the BBC.

The UK adopted its digital markets competition regime in January 2025, allowing the CMA to take targeted action against firms with "strategic market status." Google was designated last year. The CMA has given Google six months to implement fair ranking requirements and three months for data portability. The UK measures are, in some respects, more prescriptive than the EU's — they require explicit data portability for users, not just data sharing with rival search engines. The CMA is also investigating whether Google should be forced to link to rival search platforms in the UK, with a final decision expected in October 2026, according to BBC reporting. Proposed remedies include "choice screens" for users to access different search providers.

This creates a pincer movement. Google now faces coordinated but distinct regulatory regimes on both sides of the English Channel, each with the power to demand structural changes to how its search results operate. The company cannot comply with one and ignore the other. The cumulative compliance burden — and the commercial consequences of opening search data and AI features to rivals across both markets — is the real lever. Google's own head of competition, Oliver Bethell, has argued that the DMA has "materially degraded user experiences" since coming into effect, with consumers completing 50 percent more searches to reach destinations and businesses reporting 30 percent less traffic, as cited by CSIS.

Diplomat View

The EU's Google crackdown is not primarily about search results. It is about data as a moat — and the Commission has concluded that the only way to dislodge a gatekeeper whose advantage compounds with every query is to force it to pipe that advantage to competitors. The 2017 Shopping remedy proved that behavioral display requirements do not work; Google complied with the letter and captured the mechanism through an auction system that routed less than one percent of traffic to rivals. The DMA's data-sharing remedy is the structural response, and it is the first time a regulator has attempted to turn a dominant firm's data exhaust into a utility.

The strategic irony is sharp. Brussels is using competition law to force an American company to subsidize the data needs of other American companies — OpenAI, Perplexity, Microsoft — that are already attacking Google's core business from the AI side. European search rivals like Ecosia and Qwant will get access on preferential pricing terms, but they lack the engineering capacity to turn raw query and click data into competitive search or AI products at scale. The remedy that is supposed to rebalance the market toward European innovation may instead accelerate American AI consolidation of the search layer.

Google's most likely response is a two-front legal war: challenging the data-sharing measures before the General Court on privacy and proportionality grounds, while complying minimally with the self-preferencing ruling to avoid daily penalties. The Court of Justice's September 2024 ruling upholding the Shopping fine and the July 2026 ruling upholding the €4.1 billion Android fine suggest the judiciary will defer to the Commission, but the DMA's novel ex-ante framework is untested at the appellate level. Google will argue that sharing search data — even anonymized — raises re-identification risks and that the scope of mandated data sharing exceeds what Article 6(11) permits. That challenge will take two to three years to resolve, during which Google will be operating under the Commission's measures.

What to watch:

• July 27, 2026 — Target date for the Commission's final binding decision on search data-sharing measures. If the preliminary findings hold, Google must begin building the data-sharing API within months.
• 60 days after the self-preferencing ruling — Google's deadline to propose compliance changes or face daily penalties. Ribera's ad tech warning about structural remedies sets a precedent for escalation.
• October 2026 — UK CMA final decision on whether Google must link to rival search platforms, including potential "choice screens" for UK users.

The forecast hinges on whether the data-sharing API actually delivers usable, non-degraded data to rivals. If Google complies with the letter — providing data with maximum permitted latency, aggressive suppression, and minimal granularity — the remedy replicates the Shopping failure. If the Commission enforces the parity principle in practice — data as useful as what Google uses internally — the search market's architecture changes fundamentally. The difference between those outcomes is not legal; it is technical and political, and it will be fought over API specifications, not court judgments.