On March 19th, OpenAI announced it was acquiring Astral — the startup behind uv, Ruff, and ty, three of the best Python tooling projects to come out of the open-source world in years. The whole Astral team is joining Codex. A few days before that, the Wall Street Journal reported that OpenAI is merging ChatGPT, Codex, and the Atlas browser into a single desktop superapp. That's six acquisitions in 2026 alone — nearly matching everything they did in all of 2025.
Something shifted. And if you're watching this closely, it's pretty obvious what.
The Numbers Behind the Panic
OpenAI's own applications chief, Fidji Simo, sent an internal memo declaring a "code red" over Anthropic. Why? Because Anthropic now captures roughly 73% of first-time enterprise AI spending. Developers are choosing Claude. Companies onboarding to AI for the first time are choosing Claude. That's the market that matters most — the people who haven't locked in yet.
OpenAI built the category. They got everyone talking about AI. And now they're watching a competitor walk in and convert the leads. That's a nightmare scenario for any company, and it explains why the product strategy has lurched from "ship everything separately" to "merge everything immediately."
Simo's memo reportedly said that spreading team energy across too many standalone apps was slowing momentum and hurting quality control. Sora launched standalone and then quietly faded. The Codex CLI existed but didn't connect to anything. The Atlas browser was a separate product nobody quite understood. In hindsight, it all looks like feature sprawl without a strategy.
What the Superapp Actually Is
The plan: combine ChatGPT (the conversational interface), Codex (the coding agent), and Atlas (an AI-native browser) into one desktop application. Primary audience is developers and business users. The pitch is end-to-end agentic workflows — where the chat, the code editor, and the browser all share context and feed each other.
That's actually a coherent vision. If you're building something, you want your AI assistant to know what you're looking at in the browser, what code you're working on, and what you asked it five minutes ago. Right now you're copy-pasting across three different windows. A unified shell could fix that.
But here's where I'll pump the brakes a little: ChatGPT will apparently still exist as a separate app. So it's not a complete consolidation — it's a second product aimed at a different audience. The superapp targets power users; ChatGPT stays for everyone else. That's fine, but it also means they're still maintaining two codebases, two user experiences, two roadmaps. The "too many apps" problem doesn't fully go away.
What Astral Actually Means
The Astral acquisition is the one that stings a little. Not because OpenAI acquiring a company is unusual — they do that now — but because of what Astral was.
uv is the fastest Python package installer that exists. Ruff replaced flake8 and pylint for most people who tried it. ty is their new type checker, still in early stages. All three are open-source, and all three have been adopted by a significant chunk of the Python ecosystem. Developers trust them because they're fast, they're maintained well, and they're not owned by anyone with an agenda.
That last part changes now. OpenAI owns them.
The founding team — led by Charlie Marsh — joins OpenAI's Codex team. The projects will ostensibly stay open-source. But the question that's already blowing up on Hacker News is the obvious one: as they gobble up the tools that developers rely on, how viable is it that those tools stay truly open? Or does open-source become a distribution strategy for pulling devs deeper into the Codex ecosystem?
I don't think OpenAI is going to immediately close these tools off. That would be reputational suicide in the developer community. But the concern isn't that they'll kill them outright — it's that they'll slowly steer them. Prioritize features that benefit Codex. Deprioritize contributions that don't serve the commercial roadmap. Use the trust built by Charlie Marsh's team and gradually redirect it toward OpenAI's interests.
Maybe that doesn't happen. But it's the exact pattern we've seen with other open-source projects acquired by large companies. The initial promise holds, then the incentives creep in.
The Acquisition Velocity Is the Real Signal
Six acquisitions in under three months. Let's run through them quickly:
- Sky — a Mac AI interface, now foundational to ChatGPT's desktop presence
- Astral — Python tooling (
uv,Ruff,ty), absorbed into Codex - Promptfoo — AI red-teaming and evaluation framework
- Three more that haven't gotten as much coverage
This isn't a company that's carefully selecting strategic targets. This is a company buying optionality at scale. They're acquiring teams, technologies, and brand trust across the developer tooling stack — not because they have a perfectly laid-out plan, but because they need to close the gap with Anthropic fast.
When companies acquire this quickly, integration becomes the hard part. Absorbing six teams with six different cultures and six different codebases in three months almost never goes cleanly. The products you're building on top of them are being assembled from parts that haven't been tested together. That's how you end up with the very quality control problems Simo's memo was trying to solve.
"Spreading team energy across too many standalone apps was slowing momentum and hurting quality control." — Fidji Simo, OpenAI Applications Chief, internal memo (via WSJ)
That memo describes the exact problem that an acquisition spree accelerates. It's a bit of a contradiction. You fix fragmentation by buying more things and merging them together faster? The logic only works if the integration is genuinely tight from day one — and that's hard to pull off when you're moving this fast.
Where This Actually Goes
Here's my read: the superapp is real and it'll ship. It'll probably be impressive when it does. The Astral acquisition will make Codex meaningfully better for Python developers specifically. The consolidation will help OpenAI tell a cleaner story to enterprise buyers — "it's all one thing" is an easier pitch than "here's six different products."
But the 73% enterprise number isn't going away because OpenAI merged their apps. That number is about trust, capability, and developer experience over time. Anthropic built that lead by being really good at the core thing — the model, the API, the documentation, the reliability. You can't acquisition-spree your way to that. You have to earn it.
What I'm watching: whether the open-source projects stay genuinely open once the initial goodwill period passes. Whether the superapp actually delivers on the "shared context across everything" promise, or whether it's just a tabbed interface with a new name. And whether Anthropic responds with anything, or just keeps quietly winning market share while OpenAI makes headlines.
The race is real. It's legitimately competitive in a way it wasn't a year ago. But right now, OpenAI's moves feel reactive. They're responding to Anthropic's lead, not setting the pace. That might change — they have the resources and the name recognition. But buying Astral and merging some apps isn't the same as building what people actually want to use every day.
We'll see if the superapp changes that. I want it to. Better tools are good for everyone. But the internal memo that started all this wasn't "we have a great vision." It was "we're losing."
That's where OpenAI is right now.