On May 18, Anthropic announced it had acquired a company called Stainless. The price wasn't disclosed, but The Information had reported a week earlier that the deal was worth more than $300 million.

There's a reason this is more than routine M&A news. Stainless is the company that has built all of Anthropic's official SDKs — and, at the same time, the SDKs of OpenAI, Google, Cloudflare, Replicate, and Runway. In other words, Anthropic just bought the core infrastructure its own competitors were running on.

And alongside the acquisition, Anthropic made one thing clear: it is shutting down Stainless's hosted products and SDK generators entirely. Existing customers can keep using the SDKs already generated, but they will no longer get new builds or automatic updates. From now on, the tool works only inside Anthropic.

It's a signal that competition in the AI industry has moved one layer deeper.

What Is an SDK, Anyway?

To understand this news, you first need to know what an SDK is.

If an API is a restaurant's menu, an SDK is the manual that tells you how to order there. When an AI company opens up its API, developers need a way to call it from their own programs. But hand-crafting HTTP requests, parsing JSON, and handling authentication every single time is tedious. So companies ship a library that makes all of this easier. That's an SDK.

OpenAI's Python library, Anthropic's TypeScript library, Google's Java library — these are all SDKs. They're the bridge that lets a developer pull a model into their code with a single line: `import openai` or `import anthropic`.

What Stainless did is interesting. It built a tool that takes an API specification as input and automatically generates SDKs across languages — Python, TypeScript, Kotlin, Go, Java, and more. When the API changes, the SDKs update automatically, so no one has to maintain them by hand.

This matters because, for an AI company, SDK maintenance is a headache. Models ship updates every week, and if the SDK can't keep up, developers can't use the new features. That's why OpenAI, Anthropic, and Google all used Stainless. There was simply no better tool for automatically generating and maintaining their SDKs.

The Battlefield Just Got One Layer Deeper

So far, competition among AI companies has gone through three stages.

Stage one: model quality. Who builds the smarter model? GPT-4 versus Claude 3 versus Gemini. Companies competed on benchmark scores, reasoning, and coding ability.

Stage two: developer experience. Who builds the better tools? Claude Code, Codex, Cursor. This was the stage of pulling developers in through CLIs and IDE integrations.

Stage three: infrastructure ownership. Who controls the infrastructure those tools run on? The Stainless acquisition is the signal that this stage has begun.

Each stage determines the one above it. A great model goes unused if the developer experience is bad. A great developer experience can be cut off at any moment if the infrastructure sits in a competitor's hands. So Anthropic simply bought the infrastructure it had been depending on.

Whichever SDK Is the Default Wins

Picture a new developer trying an AI model for the first time. The first thing a search turns up is the official SDK. `pip install anthropic`, `pip install openai`, `pip install google-generativeai`. That one command determines which AI company's client gets installed first.

Here's the decisive part. Once you start writing code against one SDK, switching to another isn't easy. The function signatures differ, the error handling differs, the streaming differs. Which SDK a developer learns first ends up determining which model they use from then on. It's a lock-in strategy.

Zoom out, and "which SDK is sitting in the production code" determines "which model gets wired into that company's revenue." An SDK isn't just a client library — it's where the revenue stream begins.

This makes Anthropic's reasoning for buying Stainless clear. If it can raise the quality of its own SDKs faster than its competitors can — while making sure those competitors no longer have access to the same tool — the friction of choosing Claude goes down, and the friction of using any other model goes up. The SDK itself becomes a competitive advantage.

The Next Move After Overtaking OpenAI in the Enterprise

What makes this acquisition even more meaningful is the timing. Last month, reports said Anthropic had overtaken OpenAI in enterprise customer count — the first reversal in a domain where OpenAI had always held the lead.

Right after that milestone, Anthropic's next move was buying Stainless. One move, two effects: cementing the lead it had just taken, and weakening its competitors' infrastructure at the same time.

With Stainless's hosted products shut down, OpenAI and Google have to either go back to maintaining their SDKs by hand or find another tool. In the short term, that means higher costs and slower shipping. It's not a matter of days or weeks — but given that AI models update weekly, the accumulated gap can widen into something meaningful.

Anthropic's Vertical Integration Play

The Stainless acquisition signals that Anthropic's vertical integration strategy is now in full swing. The company is building a stack it controls from top to bottom.

The model layer. Claude Opus 4.7, Sonnet 4.6, Haiku 4.5. Anthropic builds the models itself.

The developer tools layer. Claude Code and Claude Desktop. It builds the CLI and IDE tooling itself.

The SDK layer. With Stainless, it now owns the SDK generation tooling outright.

The consulting layer. The $1.5 billion joint venture with Blackstone and Goldman Sachs announced on May 4 — a consulting organization that installs Claude directly into private equity portfolio companies.

The infrastructure layer. Cloud infrastructure secured through its partnerships with Amazon and Google.

All five layers are coming under Anthropic's control, in a structure where an advantage at one layer feeds the next. A good model makes for good tools; good tools make the SDK feel natural; a natural SDK makes the consulting faster; faster consulting grows revenue.

OpenAI Won't Sit Still

Anthropic's moves put the same pressure on OpenAI, because they signal that model quality alone is no longer enough.

On May 4, OpenAI announced a $10 billion deployment company — an aggressive bid to lock down enterprise channels, going as far as offering private equity backers a 17.5% guaranteed return. The Stainless episode is likely to be another jolt. OpenAI could harden its own SDK infrastructure, acquire another tooling company, or build new tools in-house. One way or another, a response is coming.

Google faces similar pressure to strengthen its own SDK infrastructure. And smaller players like Cloudflare, Replicate, and Runway have suddenly lost a core tool. A market has just opened for someone to build the tool that fills that gap.

What Korean Developers and Companies Should Watch

This shift has direct implications for developers and companies in Korea.

Choosing an SDK is choosing a model. If you're starting a project that adds AI features, which SDK you adopt determines your dependency story for the next six months. The question isn't just "which model is best" — it's "which SDK is going to be embedded in our code."

Diversifying dependencies matters more than ever. Depend on a single company's SDK and you're exposed to that company's policy changes. Use Claude, GPT, and Gemini for different kinds of work, and abstract the interface so swapping out an SDK stays easy.

Korean AI companies need to invest in SDK quality, too. Several companies in Korea are building their own AI models. Competing on model quality alone means losing to big tech. SDKs specialized for Korean-language processing, and tooling fitted to Korean development environments, can be the differentiator.

Developer experience is revenue. Korean companies selling APIs or SaaS to businesses should take the same lesson. Building a good product isn't enough. Polishing your SDK and documentation so a developer can make their first successful call within five minutes — that's where revenue starts.

Who Owns the Developer?

The real message of this acquisition is this: AI companies are no longer competing on "who builds the smarter model." They're competing on "who owns the developer's path of least resistance."

The tools developers use every day, the libraries they call every day, the docs they read every day — when all of that sits under one company's control, that company's model ends up embedded in every product those developers build. Revenue is the natural consequence.

Stainless was one piece of that control. It looked small, but it was core. Anthropic paid more than $300 million for that piece because it understands the size of the compounding effect that one piece produces.

From now on, when you read acquisition news from AI companies, watch for one thing: which tool in a developer's daily routine does this acquisition control? That tool is becoming a more important indicator than the model itself.

From competing on model quality, to competing on developer experience, to competing on infrastructure ownership — the battlefield keeps sinking one layer deeper. And the deeper it goes, the harder it becomes for small companies to catch up. The consolidation game among the giants has begun in earnest.

Which SDK becomes the default determines which model goes into production. And who owns that SDK determines the next five years of the AI market.