France's Antitrust Authority Bombshell Report: OpenAI, Google, Anthropic Control 84% of AI Agent Market

This opinion divides the AI agent market into two tiers. The first tier is the 'Dominant Tier,' composed of OpenAI, Google, and Anthropic. OpenAI dominates with GPT-5.6's full rollout across ChatGPT, Codex, and the API, plus the new ChatGPT Work product. Google gained a massive user base by setting Gemini 3.5 Flash as the default model in AI Mode globally. Anthropic rapidly captured market share through Claude Opus 5's launch, with a pricing strategy roughly half that of Claude Fable 5. The second tier is the 'Challenger Tier,' including companies like DeepSeek, Moonshot AI, and Mistral. DeepSeek V4 reached general availability with a 1 million token context window; Moonshot AI's Kimi K3 attracted attention with 2.8 trillion parameters and pricing at $15 per million tokens; Mistral shipped a robotics-focused model as its valuation approached $23 billion.

84% market concentration is rare in the tech industry. For comparison, in the search engine market, although Google dominates, competitors like Bing and DuckDuckGo still maintain certain shares; in the social media market, Meta's Facebook faces intense competition from TikTok, Twitter/X, and others. But in the AI agent market, the top three companies together control over three-quarters of the market — this concentration exceeds even traditional telecommunications and energy industries. France's Competition Authority noted in the opinion that this highly concentrated market structure may lead to multiple problems: first, slowed innovation — when a few companies control most of the market, they may lack sufficient competitive pressure to continuously innovate; second, price increases — monopolists may use market power to raise prices, harming consumer interests; third, data concentration — AI agents need to process large amounts of user data, and market concentration means data is also concentrated in a few companies' hands, increasing risks of data abuse and privacy breaches.

Notably, this opinion's release comes against the backdrop of the EU's increasingly strengthened regulatory intensity toward tech giants. The Digital Markets Act (DMA) and Digital Services Act (DSA), which took effect in 2024, have already imposed strict regulatory requirements on companies like Google, Meta, and Apple. Now, with AI technology's rapid development, the EU's regulatory reach is extending into the AI field. France's Competition Authority's opinion may provide a basis for EU-level AI antitrust action. In fact, the EU is already discussing competition policies targeting the AI market based on the AI Act, and France's report provides empirical foundation for this discussion. If the EU decides to conduct antitrust investigations into the AI agent market, OpenAI, Google, and Anthropic may face regulatory pressure similar to what they face in search engines and social media.

From a global perspective, AI agent market concentration issues are not limited to Europe. In the US, the Federal Trade Commission (FTC) has begun paying attention to the AI industry's competitive situation; in the UK, the Competition and Markets Authority (CMA) is also conducting similar investigations. It is foreseeable that in the coming years, the AI industry will face antitrust scrutiny from multiple jurisdictions globally. For OpenAI, Google, and Anthropic, this scrutiny may force them to change business behaviors — for example, opening API interfaces, lowering prices, increasing interoperability. For challenger companies, regulatory intervention may create more market opportunities — for example, through forced ecosystem opening, making it easier for new entrants to acquire users. For ordinary users, antitrust action may bring lower costs, more choices, and better services.