The platform trap: why Anthropic’s strategy mirrors Microsoft’s past

What this means for makers and artists

Creators and developers are facing a new reality where the very tools they rely on to build their businesses are being subtly held back. Anthropic is quietly limiting the power of its latest models for anyone attempting to train rival software, while simultaneously launching products that directly compete with its own user base. This shift forces makers to question whether they are building on a foundation of partnership or a ladder of their own making.

The silent throttling of competitor tools

Anthropic recently released Claude Fable 5, a version of its Mythos model, but implemented a hidden restriction. The company dialed back performance specifically when customers used the model to train their own AI software or hardware. Following public criticism, Anthropic reversed course slightly, promising at least to notify users when a weaker version of the model is being served.

Industry observers see this as a precursor to a wider issue. Martin Casado, general partner at Andreessen Horowitz, warned on X that it is only a matter of time before only model creators retain access to the most powerful systems. He noted that everyone else would be left with smaller, distilled versions. Casado is also an investor and board member at Cursor, a coding startup that is both an Anthropic customer and a direct competitor to Claude Code.

While Anthropic officially states these limits are to prevent foreign adversaries or rival labs from advancing their technology, developers fear the restrictions could affect basic features for their own applications. The suspected motive remains clear: Anthropic intends to reserve the best technology for its own competing products.

Partners becoming rivals

This throttling is merely one part of a broader pattern where model providers and their customers find themselves in direct competition. Before launching its AI design tool, Claude Design, Anthropic recruited companies like Figma and Canva as partners. Immediately prior to the release, the company expanded the feature set so aggressively that the new product competed head-to-head with those partners’ existing offerings.

The result was swift: Figma withdrew from the partnership, and Anthropic’s Chief Product Officer, Mike Krieger, resigned from Figma’s board. Dylan Field, CEO of Figma, stated at a Sequoia Capital event that Anthropic had been “not consistently candid in their communications.” This was not an isolated incident; a year earlier, Anthropic launched Claude Code, an application that threatened customers like Cursor. Eventually, Claude Code surpassed both Cursor and Microsoft’s GitHub Copilot in revenue.

Quinn Slack, CEO of coding startup Amp and a long-time Anthropic customer, summed up the sentiment: “If there’s just one winner in the AI model race, they can push around and stomp their customers.”

Revenue surge and historical echoes

Behind these aggressive moves lies an unprecedented financial boom. Annualised revenue grew fivefold in just five months, reaching nearly $50 billion, a figure that overtakes OpenAI. Like OpenAI, which offers Codex to many of its own customers, Anthropic is deepening its market position. Together, the two companies are growing faster than the next 32 large AI startups combined.

Both firms are strengthening their market dominance through subsidiaries, such as OpenAI’s DeployCo. This dynamic closely mirrors earlier antitrust cases involving Microsoft and Google, which leveraged their platform dominance to favour their own applications and squeeze out partners. Both were subsequently found to have engaged in illegal monopolistic practices.

Key takeaways

• Anthropic is restricting access to its most powerful models for users building competing AI tools, a strategy that has already prompted partial reversals following backlash.
• The company’s rapid expansion into design and coding tools has alienated key partners like Figma, creating a conflict of interest between customer success and corporate growth.
• With revenue surging to nearly $50 billion, Anthropic and OpenAI are prioritising their own product ecosystems, echoing the anticompetitive tactics that once led to legal battles against Microsoft and Google.