OpenAI’s Astral deal reveals hidden funding rounds, investor windfall: AI researcher Simon Willison

When OpenAI on March 19 announced that it would acquire Astral, the startup behind three of the Python ecosystem's most used open-source developer tools, the deal was framed as an acceleration of its Codex AI coding platform. But independent AI researcher Simon Willison says the announcement raises more questions than it answers and the most important ones involve money that was never publicly disclosed.

Willison flagged that Astral completed both a Series A and a Series B funding round without ever making either public. "As far as I can tell neither the Series A nor the Series B were previously announced," he wrote in a post.

He added that investors including Accel and Andreessen Horowitz now stand to benefit as their stakes are effectively converted into equity in OpenAI, which reportedly could go public as soon as the end of this year. Willison also raised the question of how much influence those investors may have had in pushing the deal forward.

Also read: OpenAI to acquire Astral to boost Codex and expand AI-driven software development tools

"Those investors presumably now get to exchange their stake in Astral for a piece of OpenAI. I wonder how much influence they had on Astral's decision to sell," Willison said. 

A load-bearing acquisition

At the centre of the deal is uv, Astral's Python environment management tool, which Willison describes as the most consequential of the company's three products and possibly the most strategically valuable asset OpenAI has just acquired.

"uv is the big one," Willison wrote. Released in early 2024, the tool was downloaded more than 127 million times last month alone, according to PyPI Stats. Python's historically chaotic dependency management, satirised in a well-known XKCD comic depicting an incomprehensible tangle of overlapping environments, has been substantially resolved by uv, Willison argues. "Switch from Python to uv run and most of these problems go away," he wrote.

Also read: OpenAI plans to build a desktop 'Superapp' combining ChatGPT, Codex app, and Browser

Astral's two other tools, Ruff, a Python linter and formatter, and ty, a type checker, are popular but, in Willison's assessment, not load-bearing in the same structural sense as uv. He acknowledged their relevance to AI coding agents, noting that fast linting and type-checking tools can help improve output quality, but expressed scepticism that deep integration into Codex itself would be transformative. "I'm not convinced that integrating them into the coding agent itself as opposed to telling it when to run them will make a meaningful difference," he wrote, while allowing that he may lack the imagination to see it.

Talent or product...or both?

Willison also questioned whether the strategic logic of the acquisition is primarily about Astral's tools or the engineering talent behind them. He pointed to BurntSushi, Astral's engineer known for the Rust regex engine, ripgrep, and the jiff time library, as someone whose hire alone could justify the transaction.

He cautioned, however, that deals framed as product-plus-talent acquisitions can quietly become talent-only acquisitions once the acquirer has what it actually wants.

"I know from past experience that a product+talent acquisition can turn into a talent-only acquisition later on," Willison said. 

Infrastructure as a competitive weapon

The acquisition also comes amid rising competition between OpenAI and Anthropic in AI-assisted coding.

“The competition between Anthropic’s Claude Code and OpenAI’s Codex is fierce,” Willison said, adding that coding agents have become “almost-indispensable tools for software development.”

Also read: Anthropic expands Claude Cowork with new enterprise plugins for finance, HR, engineering and others

illison noted that Anthropic acquired the Bun JavaScript runtime in December 2025, a move with structural similarities to the Astral acquisition, and that Claude Code's performance has improved significantly since Bun's creator, Jarred Sumner, joined the company.

Both deals reflect a larger pattern. AI labs are acquiring open-source infrastructure that is deeply embedded in developer workflows, then using that ownership to accelerate their own coding products. But Willison flagged a risk that goes beyond competitive dynamics.

"One bad version of this deal would be if OpenAI start using their ownership of uv as leverage in their competition with Anthropic," he wrote.