This website uses cookies

Read our Privacy policy and Terms of use for more information.

Judge Mehta ruled Google a monopolist in August 2024. On September 2, 2025, he issued the first remedies opinion. On December 5, he finalized the judgment. On April 13, 2026, the DOJ filed its appeal seeking stronger remedies.

Nearly two years of procedure to reach a point almost no European SME is watching.

Yet it concerns every company that relies on a search engine to be found, every company that pays to appear at the top, every company that has started integrating Gemini into its workflows. It concerns anyone who has built parts of their business on top of an infrastructure that a court has just declared illegally monopolized.

It's worth understanding what actually happened. And more importantly, what didn't.

The verdict everyone missed

The United States v. Google case goes back to 2020. The charge: Google paid billions of dollars to Apple, Samsung, and other manufacturers to be the default search engine on their devices, effectively blocking competition. In 2024 the judge ruled that yes, this behavior violated Section 2 of the Sherman Act.

The interesting question was a different one: what to do about it?

The DOJ wanted two heavy things. First: forced divestiture of Chrome, to break the monopoly structurally. Second: mandatory sharing of the search index with qualified competitors, so that Bing, DuckDuckGo, and any future challenger could start from a comparable data foundation.

The first remedies opinion came in September 2025. The final judgment was entered on December 5. The result was a half-compromise that neither side accepted.

What the DOJ got:

  • Six-year ban on exclusive distribution contracts for Google Search, Chrome, Google Assistant, and the Gemini app

  • Prohibition on conditioning access, payments, or favorable terms for any Google product on use of another Google product (the "conditioning rule")

  • Obligation to share portions of the search index with qualified competitors

  • Obligation to share aggregated user-interaction data (queries + results + clicks)

  • Creation of a technical oversight committee with expertise in AI, economics, and data privacy

  • One-year cap on default placement agreements (no more multi-year contracts with Apple)

What it didn't get:

  • Divestiture of Chrome

  • Divestiture of Android

  • Mandatory choice screens

  • Data sharing with anyone (only "qualified" competitors)

  • Self-preferencing prohibitions (the court held that "could prove anticompetitive" wasn't sufficient basis)

Google filed its appeal on January 16, 2026, challenging the data-sharing requirements and technical committee oversight. The DOJ filed a cross-appeal on February 3 seeking stronger remedies including forced divestitures. On April 13 the DOJ formalized its appeal.

Appeals court hearings are scheduled for late 2026 or early 2027. During the appeal, the remedies remain in force — or perhaps not, since Google has also asked for a stay pending the ruling.

The detail that changes everything

There's a point that mainstream coverage has almost entirely ignored.

The remedies apply explicitly to generative AI.

Not just to traditional search. Not just to Chrome and Assistant. The final judgment — issued three months after the initial remedies opinion — covers any "application, software, service, feature, tool, functionality, or product" that involves generative AI or large language models. Not just Gemini. Anything Google builds in the GenAI space.

The mechanism is specific. The court adopted what Winston & Strawn describes as a "Google shall not condition" rule: Google cannot make access, payments, or favorable terms for one product contingent on using another Google product, setting it as the default, or sidelining a competitor. Translated: if you want Google Search on your device, Google can't require you to also take Gemini. If you want Play Store access, you can't be forced to bundle an AI assistant.

In the December finalization, Judge Mehta was blunt. Adopting Google's narrower proposal, he wrote, would be "self-defeating." Google cannot be permitted to "replay its illegal conduct with its GenAI products."

It's the first time a U.S. court has imposed explicitly forward-looking antitrust remedies on generative AI.

Not "after you've monopolized this market too we'll fine you," but "you can't use the monopoly you already have to win the next market."

The ruling has limits. It opens distribution channels but doesn't touch the underlying data advantage. Google still runs 14 billion searches a day, generating real-time feedback loops that no static data snapshot can replicate. As Brookings observed, the remedy addresses the data Google accumulated through its monopoly, but not the ongoing advantage it maintains by continuing to operate it.

The court bet on distribution openness being enough. Whether it is remains an open question — and the one the appeal will likely center on.

What this means in practical terms

Three medium-term effects worth thinking about.

First effect: the end of exclusive contracts for consumer AI.

Until September 2025, Google could theoretically pay a smartphone manufacturer to preinstall Gemini as the only AI assistant. It could pay a browser to integrate Gemini as the only chat option. It could condition Android licenses on Gemini bundling.

Not anymore. For six years.

This opens space for challengers who previously had to compete against a distribution wall. It doesn't make them automatically winners — but it at least lets them exist in the consumer market without paying structural tolls.

Second effect: access to search data.

Qualified competitors will be able to obtain portions of Google's index and aggregated interaction data. For an SME this isn't directly useful — no SME will become a qualified competitor to Google. But indirectly it changes the market.

An alternative search engine that today returns worse results than Google because it has less data could tomorrow return better results — not necessarily comparable, but closer than the current gap allows. As ProMarket noted, the data-sharing remedy provides static snapshots, not access to real-time feedback loops. Still, the direction matters. Investing in organic positioning on platforms that aren't Google starts to make technical sense. Not for ideological reasons. For coverage.

Third effect, the subtlest: the legitimacy of behavioral remedies.

Mehta rejected the structural path (no Chrome divestiture) in favor of the behavioral path (conduct obligations + technical oversight). If the appeal upholds this approach, it sets a precedent. The next time an antitrust authority — American or European — has to address a similar case on a dominant AI infrastructure, it will have a model to follow.

And the next cases are coming. The FTC appealed the November 2025 ruling that cleared Meta, filing its notice on January 20, 2026 — doubling down on the "killer acquisition" theory for Instagram and WhatsApp. The European Commission opened a formal antitrust investigation into Google's use of publisher content for AI training. The DOJ has launched two separate antitrust probes into Nvidia for GPU bundling practices and acquisition conduct.

The playbook emerging from the Google case will become the foundation for all of them.

The European side of the question

The EU AI Act regulates AI use on European territory. European antitrust regulates the market conduct of platforms.

These are two separate planes that rarely meet.

But the question the Google case raises — what structural obligations should dominant AI platforms face to prevent replicating existing monopolies in a new market — is exactly the question Europe is actively pursuing, case by case, platform by platform.

The Commission has tools. The Digital Markets Act identifies "gatekeepers" and imposes interoperability obligations. The Digital Services Act imposes transparency requirements. The AI Act imposes governance obligations.

And unlike in 2024, these tools are now being applied specifically to AI bundling.

On December 4, 2025, the Commission opened formal antitrust proceedings against Meta for blocking third-party AI assistants from WhatsApp Business. By February 2026 it had issued a Statement of Objections. On April 15, 2026, it issued a Supplementary Statement of Objections, finding that Meta's fee model for AI access was functionally equivalent to the original ban — and announced its intent to impose interim measures. ChatGPT, Copilot, and Perplexity all lost WhatsApp access in the EEA outside Italy.

On December 9, 2025, the Commission opened a formal antitrust investigation into Google's use of online content to train its AI models and generate AI-powered overviews in search results, without compensating publishers or giving them the option to opt out.

The Commission continues its investigation into Microsoft's integration of Teams with Microsoft 365 — a case that began with Slack's 2020 complaint. Microsoft preemptively unbundled Teams in Europe in 2023, but the Commission determined that this didn't fully address the competitive harm. The UK's Competition and Markets Authority is pursuing a parallel investigation into Microsoft's cloud licensing terms that favor Azure over AWS and Google Cloud.

The pattern is clear: Europe is building its own version of the conditioning rule. Not through a single landmark ruling, but through a convergence of proceedings that all address the same question: can a dominant platform tie access to its core product to the use of its AI product?

And then there's the sovereignty push.

What the Commission is doing through competition law, France is doing through procurement policy — and at a scale that has no precedent in Western Europe.

On April 8, 2026, France's Interministerial Digital Directorate (DINUM) announced that it is migrating all government workstations from Windows to Linux and ordered every ministry to formalize a plan to eliminate extra-European digital dependencies by autumn 2026. The directive covers operating systems, collaborative tools, cloud infrastructure, and AI platforms.

This isn't a pilot. It follows France's January 2026 mandate to replace Microsoft Teams, Zoom, Webex, and GoTo Meeting with Visio — a sovereign, open-source videoconferencing platform built on Jitsi, hosted on Outscale servers (a Dassault Systèmes subsidiary), and certified by France's national security agency ANSSI. Already tested by 40,000 users, it's being deployed to 2.5 million civil servants by 2027. The government estimates savings of €1 million per year for every 100,000 users migrated.

The broader stack is already in place: La Suite Numérique includes Tchap for messaging, Docs for collaboration, Grist for spreadsheets, Nextcloud-based Fichiers for storage, Open-Xchange for email, and France Transfert for large files. All hosted on EU infrastructure. All interoperable by design.

France isn't alone. Schleswig-Holstein, in northern Germany, has migrated more then 40,000 government workstations to LibreOffice, replacing Microsoft Exchange with Open-Xchange and Thunderbird. The state projects €15 million in annual savings from 2026. Denmark's Ministry of Digital Affairs made a similar switch. The Austrian military is reported to be doing the same.

As David Amiel, France's Minister of Public Action, put it: "We must regain control of our digital destiny. We can no longer accept a situation in which we lack control over our data and infrastructure while remaining dependent on decisions made by foreign companies."

For an SME, this matters for two reasons.

1. Procurement is shifting under you. If your clients include European public administrations, the toolchains they'll accept, the interoperability standards they'll require, and the data residency constraints they'll impose are all changing.

2. Sovereignty and antitrust are the same story. The sovereignty push isn't a separate push from the antitrust ruling. It's the same logic expressed through a different instrument.

Courts break the bundling. Governments replace the stack.

What an SME should do now

Three operational considerations.

1. Diversify dependency on Google channels.

If 60% of your organic traffic comes from Google and 40% of your ad spend goes to Google Ads, you're exposed to any change in platform logic. The next twelve to twenty-four months will see Google under pressure on multiple fronts: the appeal, the remedies in force, AI integrations that need to be redesigned to comply with bundling prohibitions. Their attention will be elsewhere. Your KPIs could oscillate for reasons that have nothing to do with your work.

You don't need to abandon Google. You need to avoid a single point of failure.

2. Actively evaluate AI alternatives.

For years the choice of AI stack has been implicit: if you're on Google Workspace, you use Gemini; if you're on Microsoft 365, you use Copilot; if you need something better, you pay ChatGPT Enterprise. Native integrations were the default choice.

That logic is cracking. With bundling bans and data access opened to qualified competitors, it becomes sensible to evaluate every AI tool on its specific capabilities, not on its integration with the ecosystem you're already locked into.

This is our approach at ELECTE: we build on top of exposed APIs, not on top of closed integrations, precisely because the underlying terrain isn't stable. An SME that in 2026 builds a critical process on top of an AI assistant bundled with an office suite is signing a long-term dependency on an architecture that competition law is actively dismantling — and that procurement policy, at least in France and parts of Germany, is actively replacing.

3. Know that the rules are being written now.

The real value of the Google case, for an SME, isn't in the specific remedies. It's in the fact that it establishes a first legal architecture for treating AI platforms as regulable market infrastructures.

Proceeding

Expected decision

Google appeal hearing

Late 2026 / early 2027

FTC Meta appeal (D.C. Circuit)

Same window

European Commission proceedings (Meta, Google)

Same window

France's ministry-level migration plans

Autumn 2026

Every one of these outcomes will reshape which platforms are available, how they can be distributed, what data they must share, and what dependencies they can impose. The choices you make now about AI infrastructure will either align with where regulation is heading or need to be unwound later.

The rules aren't settled. They're in the most volatile part of their cycle.

Fabio Lauria

CEO & Founder, ELECTE

Every week we explore AI without the hype — with data, analysis and an independent perspective.

Sources

If you found this analysis useful, please share it with someone who might be interested. And if you’d like to find out how ELECTE uses AI to automate data analysis and reporting, you can find out more at electe.net.

Recommended for you