A Quiet Acquisition, A Loud Signal
Last week, a seemingly minor but deeply significant event unfolded in the AI industry: Anthropic acquired the SDK tool company Stainless and immediately cut off access for OpenAI.
If you haven't heard of Stainless, that's fine—about a quarter of developers worldwide use it daily. The company specializes in SDK wrapper tools for AI APIs. In simple terms, it makes it easier for developers to call the interfaces of various large language models.
Anthropic bought it, then told OpenAI, "No, you can't use it anymore."
On the surface, this is just a business acquisition. But when you break it down, it signals that AI industry competition has entered an entirely new phase.
From "Open Protocols" to "Acquisition Lock-in": Anthropic's Three-Step Strategy
Looking at the longer timeline, Anthropic's strategy is remarkably clear:
Step 1: Open-source the MCP protocol. The Model Context Protocol (MCP)—an open standard proposed by Anthropic—enables AI models to seamlessly connect with various tools and data sources. It's a smart move: using an open standard to build an ecosystem and positioning itself as the de facto core of the connectivity layer.
Step 2: Launch Agent Skills. Claude Skills allows developers to define an agent's behavior patterns via a SKILL.md file. This effectively establishes a de facto standard at the application layer, with a growing number of frameworks adopting it.
Step 3: Acquire Stainless. Directly purchasing an infrastructure-layer tool vendor to control the critical channel developers use to call model interfaces.
With these three steps complete, Anthropic has assembled the four-piece suite: models, interface protocols, application skill frameworks, and developer toolchains. In the AI Agent ecosystem, it is no longer just a participant—it is attempting to become the rule-maker.
The Commercial Iron Fist Beneath the "Responsible AI" Narrative
In its public messaging, Anthropic has consistently positioned itself as a "responsible AI company"—emphasizing safety, alignment, and a less aggressive approach compared to other players. This positioning has won it significant enterprise clients and regulatory trust.
However, the acquisition of Stainless and the cutoff of OpenAI reveals another side: When it comes to commercial competition, Anthropic's tactics are no gentler than anyone else's.
This isn't a criticism—commercial competition is inherently ruthless. But it serves as a reminder: in the AI industry, "responsible" is a positioning strategy, not a shield. Anthropic won't show mercy in competition just because it publishes more papers on AI safety.
In fact, precisely because it has built trust around safety, its commercial aggressiveness may be even more underestimated.
Developers Become Pawns on the Chessboard
For everyday developers, the most direct impact of this move is clear: the toolchains you use to call AI models are increasingly being controlled by a handful of companies.
Following Anthropic's acquisition of Stainless, OpenAI's APIs can no longer be accessed through this tool. This means developers using Stainless who want to run both OpenAI and Anthropic models now have to maintain two separate SDKs themselves.
The deeper issue is this: as the infrastructure layer becomes vertically integrated, developers' choices are shrinking. Today Anthropic can cut off OpenAI; tomorrow it could do the exact same thing to any other competitor.
This bears some resemblance to the Android vs. iOS ecosystem wars, but it's more extreme. In the mobile ecosystem, developers can at least publish apps on both platforms. In the AI Agent ecosystem, if you are deeply tied to one platform's toolchain, the migration costs will be significantly higher.
Is the Industry Entering the "Walled Garden" Era?
Over the past two years, the dominant narrative in AI has been openness: open-source models, open protocols, open ecosystems. But Anthropic's acquisition of Stainless could mark a turning point—when ecosystem competition reaches a critical juncture, openness will yield to control.
Google has Gemini + Google Cloud, Microsoft has OpenAI + Azure, and Apple has Apple Intelligence + a closed ecosystem. Anthropic is now walking the same path: using open protocols to attract developers, then using acquisitions to lock down critical infrastructure.
This isn't necessarily a bad thing—competition drives progress. But as developers or enterprise users, we need to stay clear-eyed: In the AI industry, there are no permanent friends, only permanent interests. Protocols that are open today could be locked into walled gardens tomorrow; tools you can freely call today might switch camps overnight due to a single acquisition.
The best strategy is to enjoy the efficiency gains AI tools bring, but never bet your entire livelihood on any single ecosystem.