Rust Rewrote Python's Toolbox. Then OpenAI Came Calling
Most Python developers never run a package manager written in pure Python anymore. The tools that install your dependencies, lint your code, and manage your environments are increasingly Rust binaries. And the company that made the biggest ones, Astral, now belongs to OpenAI.
Here's how a pair of fast Rust tools quietly carried Python's packaging ecosystem, and why an AI company wanted them.
The problem Python still had
Before uv, a modern Python setup meant juggling a chain of separate tools: pyenv for Python versions, venv (or virtualenv) for environments, pip-tools to lock dependencies, pip to install them, and pipx for running command-line tools. Each spoke its own dialect of configuration and failed in its own way.
uv: one binary to rule them
Astral launched uv in February 2024 as a drop-in replacement for pip and pip-tools — a single Rust binary, no Python required, built around what founder Charlie Marsh openly calls "a Cargo for Python" (Rust's own package manager).
The speed numbers are the headline. In Astral's own benchmarks, uv is 8–10x faster than pip with a cold cache, and 80–115x faster with a warm one, thanks to a global module cache that skips re-downloading and re-building. For daily workflows, that difference stops being a nicety and becomes the default.
Adoption followed fast. By early 2025 Marsh pegged uv at over 12.5% of all requests hitting PyPI — more than 200 million a day by some counts — a figure the python.org packaging team's own BigQuery data corroborates, with uv accounting for over a fifth of installer traffic in mid-2025 (allowed for the fact that uv eagerly prefetches metadata). By mid-2026 it was driving roughly 162 million monthly downloads and still clearing 10% of request traffic.
Ruff, Astral's linter and formatter, did the same job to fixed tools: it folded Flake8, pycodestyle, isort, Black-formatter, and more into one program. Which brings us to the third piece of the story.
Why OpenAI bought them
On March 19, 2026, Astral announced it was joining OpenAI's Codex team, with OpenAI confirming the acquisition. Terms weren't disclosed. In its announcement, OpenAI said Codex had already passed 2 million weekly active users, tripled user growth and quintupled usage since the start of the year — and that Astral's tools would live "directly in that workflow" of planning changes, running tools, and verifying results.
The motive is easy to read. AI coding agents don't just generate code; they have to check it. That requires fast, deterministic, local tools — a linter, a formatter, a type checker. Astral's suite is exactly that, and its engineering team (notably BurntSushi, of ripgrep and Rust regex fame) is among the strongest Rust shops around. It was one of the most load-bearing open-source acquisitions in recent memory: hundreds of millions of downloads per month across uv, Ruff, and the newer type checker ty.
What it means
Two converging trends. First, the "Python tooling renaissance" is really a Rust story: a systems language quietly became the foundation beneath Python's day-to-day developer experience. Second, the AI-coding boom is now pulling on that base — OpenAI wanted Python's packaging, not to run a package manager, but to give Codex the fast groundwork it needs. Both parties insist the open-source tools stay open after the deal closes.
The tools you type pip install with were always going to be rewritten. It just turned out the people who did it now work for a coding-agent company.