OpenAI's $750M Astral Deal Still Isn't Closed [2026]

OpenAI has reached an agreement to acquire Astral, the small startup behind three of the most widely used tools in the Python ecosystem: uv, Ruff, and ty. The deal, first disclosed earlier in 2026 and still working through closing conditions as of October 21, folds a company whose software touches millions of developer machines into OpenAI’s Codex organization. Secondary reporting places the price at roughly 750 million dollars, though neither company has confirmed official terms.

The timing matters. OpenAI’s Codex has grown fast this year, and the company has said plainly that it wants the agent to do more than write code: it wants Codex to plan, test, and maintain software across its entire lifecycle. Astral’s tools are the plumbing that makes that possible in the Python world, which still runs a huge share of backend services, data pipelines, and AI research code itself. Buying the plumbing, rather than renting it, is a different kind of move than the model releases and pricing tweaks that have dominated AI coding headlines this year.

Google · Preferred Sources

Don't miss new tech stories on Google

Add TrendinTech once in the Google app and our stories appear in your news suggestions.

Add Now

What OpenAI is actually buying

Astral is the company behind uv, a Python package and project manager written in Rust; Ruff, a linter and formatter that Astral says runs more than 30 times faster than older tools while staying over 99.9 percent compatible with Black’s formatting output; and ty, a type checker and language server built for speed and IDE integration. All three are open source, and all three have become default choices in a large slice of Python projects over the past two years, replacing older tools like pip, flake8, and mypy for teams that wanted faster feedback loops.

Astral’s founder and CEO, Charlie Marsh, framed the deal as a continuation rather than an exit. “Astral has always focused on building tools that transform how developers work with Python, helping them ship better software, faster. As part of Codex, we’ll continue evolving our open source tools to push the frontier of software development,” Marsh said in OpenAI’s announcement. Astral’s team echoed that message in its own post, writing that “after joining the Codex team, we’ll continue building our open source tools, explore ways they can work more smoothly with Codex, and expand our reach to think more broadly about the future of software development,” according to the company’s blog.

On OpenAI’s side, Codex lead Thibault Sottiaux made the commercial logic explicit: “Astral’s tools are used by millions of Python developers. By bringing their expertise and ecosystem to OpenAI, we’re accelerating our vision for Codex as the agent most capable of working across the entire software developer lifecycle,” he said in the same announcement. That line is the clearest statement yet of why a company built around large language models would spend hundreds of millions of dollars on a linter and a package manager.

Why a chatbot company wants a package manager

For most of the last three years, the AI coding story has been about models: which one writes better code, which one is cheaper per token, which one can hold more context. Astral sells none of that. It sells speed and reliability in the unglamorous layer underneath the model, the part of the stack that resolves dependencies, catches a type error before it reaches production, and formats code so a diff doesn’t turn into noise.

That layer turns out to matter enormously for autonomous coding agents. A model that writes plausible-looking Python is only useful if the environment around it can install the right package versions, run tests reliably, and catch type mismatches before they become runtime bugs. Technology executive and investor Sridhar Bellamkonda put it bluntly in a public comment on the deal: “Many people think that ‘AI’ is just the chat they have with an LLM, not realizing that there is a huge unseen ecosystem of layers that have to work together to help achieve results… The model is useless if the environment it operates in is broken, slow, or unreliable,” he wrote on LinkedIn.

That is effectively the bet OpenAI is making. Rather than depend on external, community-governed tools that it cannot change on its own schedule, OpenAI now owns the team that builds them. It can tune uv’s dependency resolution for agent workloads, bake ty’s type signals directly into Codex’s feedback loop, and ship Ruff’s linting rules as a first-class part of the agent’s self-correction cycle. None of that requires a new model release. It requires owning the infrastructure the model depends on.

The Anthropic angle

Coverage of the deal has repeatedly framed it as a shot aimed at Anthropic, whose Claude models and Claude Code product have built a strong reputation among professional developers, including many who work primarily in Python. Anthropic has leaned on third-party and open ecosystem tooling rather than acquiring it outright, which has generally kept its developer-tools strategy lighter and faster to adapt but also more dependent on tools it does not control.

By owning uv, Ruff, and ty outright, OpenAI gets a form of control Anthropic does not currently have over the same layer of the Python stack. If OpenAI integrates Astral’s tools tightly enough into Codex, non-Codex agents, including Claude Code, could in theory still use uv, Ruff, and ty since they remain open source. But OpenAI would control the roadmap, prioritize features that benefit Codex first, and have inside knowledge of how its own agent uses the tools before anyone else. That is a meaningful edge even without exclusivity.

A deal announced once, then again

The Astral acquisition is not brand-new news in the sense of a surprise announcement this week. OpenAI’s original statement on the deal dates back to earlier in 2026, and search interest in the acquisition spiked sharply at that time before tapering off. What has brought it back into the news cycle in mid-to-late October is a wave of follow-up coverage examining why the deal still has not closed, what it means for Codex’s roadmap, and how it stacks up against Anthropic’s developer tooling strategy. OpenAI reiterated as recently as October 18 that the agreement remains in place and that Astral’s team is expected to join the Codex organization once the transaction closes.

That gap between announcement and closing is itself a story. Months after the deal was first disclosed, OpenAI’s own materials continue to describe the transaction as pending, subject to regulatory approval and other customary closing conditions. Neither company has published a specific closing date. For a deal this size in the developer-tools space, that is a long runway, and it leaves open questions about whether regulators are scrutinizing a major AI lab’s acquisition of widely used open-source infrastructure more closely than a typical startup acquihire would receive.

What changes for Python developers, and what does not

For the millions of developers who already run uv to manage virtual environments, Ruff to lint and format their code, or ty to catch type errors, OpenAI and Astral have both said the tools will stay open source after the deal closes. That promise is the single most repeated line in both companies’ public statements, and it is clearly aimed at heading off a backlash from a developer community that has watched other beloved open-source projects get absorbed and quietly deprioritized after an acquisition.

Astral has kept shipping during the pending acquisition rather than freezing development, which is itself a signal that the team is not simply waiting out a corporate transition. Ruff advanced through its 0.16 series into 0.17.0 and then 0.17.1 in October, adding signed macOS and Windows release artifacts, expanded Python 3.15 support, and broader conda-forge platform coverage. uv moved from its 0.12 series to 0.13.0 and then a 0.13.1 patch release, continuing to track the newly released Python 3.15 as its default stable interpreter version. ty’s documentation also saw updates in the same window, describing it as an extremely fast Python type checker and language server built in Rust, with incremental analysis and IDE integration features.

What is less clear is what happens over a longer horizon once Astral’s engineers report into Codex leadership rather than running an independent company. Open-source maintainers who join large companies sometimes keep shipping at the same pace for years; other times, priorities shift toward whatever serves the parent company’s product roadmap, and community-requested features slip. Astral has not published a governance plan for how feature requests from the broader Python community will be weighed against Codex’s internal priorities after the deal closes.

Codex’s growing footprint

The acquisition lands as Codex itself has been expanding quickly. OpenAI has rolled out GPT-6.1 Sol as its newest coding-focused model, available across Codex’s local, remote, and cloud environments, and has pushed Codex context windows higher to handle larger codebases in a single session. The company has also built out dedicated Codex seats for Business and Enterprise customers with usage-based pricing rather than a flat monthly fee, a sign that OpenAI is trying to make Codex adoption easier for large engineering organizations rather than just individual subscribers.

OpenAI has estimated that Codex usage averages somewhere between 100 and 200 dollars per developer per month, though it notes that actual cost varies heavily depending on model choice, concurrency, and whether a team uses Codex’s faster processing modes. That kind of variable, consumption-based pricing is a different model from the flat per-seat subscriptions that dominate much of the developer-tools market, and it puts pressure on OpenAI to make each unit of Codex usage as efficient as possible. Faster dependency resolution, faster linting, and faster type checking all shrink the compute and time Codex burns per task, which is a direct, measurable reason to own that layer rather than call out to third-party tools.

Market reaction and the broader acquisition pattern

The Astral deal is not an isolated move. OpenAI has made a habit this year of buying smaller tooling companies to round out Codex’s capabilities rather than building every layer from scratch internally. That pattern echoes a broader shift happening across the AI industry: frontier labs that once competed almost entirely on model quality are increasingly competing on the surrounding ecosystem, the testing frameworks, dependency managers, and integration layers that determine whether an agent’s output actually works in a real codebase.

For venture-backed developer-tools startups, the Astral deal sends a signal that AI labs, flush with capital, are now credible acquirers for infrastructure-layer tools that would previously have sold to enterprise customers or stayed independent indefinitely. That changes the calculus for founders building linters, formatters, test runners, or build tools: an AI lab acquisition is now a realistic exit path, not just an enterprise sale or an IPO.

Competitive comparison: how the major AI coding agents stack up

The table below lays out how OpenAI’s Codex, after the Astral deal, compares with the two other coding agents most often cited alongside it in developer conversations. Figures reflect publicly stated positioning rather than independent lab benchmarks, since the companies involved have not published a shared, apples-to-apples test.

Coding agent	Parent company	Owned tooling layer	Pricing model	Stated strategic focus
Codex	OpenAI	uv, Ruff, ty (pending Astral close)	Usage-based credits, plus Plus/Pro/Business/Enterprise seats	Full software development lifecycle: planning, testing, maintenance
Claude Code	Anthropic	Relies on third-party and open ecosystem tools	Subscription seats tied to Claude plans	Agentic coding depth and developer workflow integration
GitHub Copilot	Microsoft / GitHub	Integrated into GitHub and VS Code tooling, no dedicated Python toolchain acquisition	Per-seat subscription across individual and enterprise tiers	IDE-native assistance and enterprise distribution through GitHub

Astral’s tools, before and after the deal

It helps to separate what each Astral tool does and where it stood heading into the acquisition. The table below summarizes the three products at the center of the deal.

Tool	Function	Written in	Recent release activity (October 2026)	Open-source status after deal
uv	Python package and project manager	Rust	Advanced through 0.13.0 to a 0.13.1 patch release; defaulted to the newly released Python 3.15	Confirmed to continue as open source
Ruff	Python linter and code formatter	Rust	Released 0.17.0 and 0.17.1, adding signed release artifacts and Python 3.15 support	Confirmed to continue as open source
ty	Python type checker and language server	Rust	Documentation updated in October describing incremental analysis and IDE integration features	Confirmed to continue as open source

Historical context: AI labs buying the shovels, not just the gold

Software companies acquiring developer-tools startups is nothing new. What is newer is an AI model company doing it specifically to make its own agent more capable, rather than to round out a product suite sold directly to developers as a standalone business. For years, the dominant strategy among AI labs was to treat the surrounding developer ecosystem, package managers, linters, CI systems, as a neutral layer they would simply plug into. Astral’s acquisition suggests that strategy is shifting, at least at OpenAI, toward treating that layer as something worth owning outright when it is central to the product experience.

It also echoes a pattern familiar from earlier eras of software history, when platform companies bought the infrastructure tools their flagship products depended on rather than risk being at the mercy of a third party’s roadmap or, worse, a competitor’s. The difference this time is the speed: Astral went from a relatively young, well-regarded open-source shop to part of one of the most valuable private companies in the world within the same calendar year, reflecting just how fast the AI coding market is consolidating around a handful of dominant platforms.

What could go wrong

The clearest risk is community trust. Open-source developers have been burned before by promises that a newly acquired project would “stay open source” only to see feature development slow, governance shift toward a corporate roadmap, or community pull requests quietly stall. Astral’s tools are embedded deeply enough in the Python ecosystem, used inside continuous integration pipelines, Docker images, and IDE extensions across a large share of professional Python shops, that any sign of OpenAI deprioritizing non-Codex use cases would likely trigger a fast and vocal reaction.

There is also regulatory uncertainty. The deal remains subject to customary closing conditions, including regulatory approval, months after it was first announced. That is a longer gap than many comparable developer-tools acquisitions take to close, and it raises the possibility that antitrust or competition authorities are looking more closely at a major AI lab absorbing widely used open infrastructure than they would at a typical startup acquisition.

Finally, there is execution risk inside OpenAI itself. Integrating a small, focused team used to running independent open-source projects into a large, fast-moving product organization built around Codex is not guaranteed to go smoothly. Astral’s engineers built their reputation on speed and narrow focus; keeping that culture intact inside a much larger company is a real test of the integration, not just the announcement.

Predictions: where this goes next

Expect OpenAI to formally close the Astral deal within the coming months once remaining regulatory and customary conditions clear, likely with a renewed public statement reaffirming the open-source commitment.
Look for Codex to ship tighter integration with uv, Ruff, and ty specifically, faster dependency resolution inside agent sandboxes, inline linting feedback during code generation, and type-aware self-correction, well before any other AI lab matches that depth.
Anthropic and GitHub Copilot’s parent Microsoft are likely to respond not with a matching acquisition necessarily, but with deeper partnerships or investments in existing open tooling providers to avoid ceding the same ground.
Other AI coding platforms will likely pursue similar acquisitions of infrastructure-layer tools, such as test runners, build systems, or CI tooling, as the market increasingly competes on reliability and speed rather than model quality alone.
Expect continued scrutiny from parts of the open-source community over governance, with pressure on OpenAI to publish a clearer long-term roadmap for how community contributions to uv, Ruff, and ty will be handled once the Astral team is fully inside Codex.

Why this matters beyond Python

Python is the immediate battleground, but the underlying lesson applies well beyond one programming language. As AI coding agents like Codex, Claude Code, and tools built on top of GitHub Copilot and Cursor compete for professional developers, the quality of the model is increasingly only part of the story. The tooling layer underneath, package management, linting, type safety, testing, determines whether an agent’s output is trustworthy enough to ship without a human rewriting half of it.

That is why a deal involving a company most casual developers have never heard of has drawn this much attention from people who watch the AI coding space closely. It is not about Astral’s revenue or user count, which have not been disclosed. It is about what the acquisition signals: that reliable, fast developer infrastructure has become a strategic asset worth owning outright, not just a dependency to tolerate. Research like the recent METR study on AI coding agents has already shown that agent-assisted development does not automatically translate into faster delivery, which makes the surrounding tooling, and whether it actually speeds up real work, matter even more than the headline model benchmarks.

It is also a reminder of how fast the ground is shifting underneath developer tools generally. Just this year, Python 3.15 shipped with its own set of performance changes that Astral’s tools had to track almost immediately, while OpenAI’s own Codex CLI has been adding new interaction features like voice control within the same stretch of weeks. Developer platforms that once updated on an annual cycle are now shipping meaningful changes monthly, and companies that cannot keep their tooling layer moving at that pace risk falling behind regardless of how good their underlying model is. OpenAI’s decision to buy rather than depend fits a broader industry trend visible in moves like Replit’s recent default-model switch between OpenAI and Anthropic, where platform-level choices about which AI provider to build around are increasingly treated as strategic bets rather than simple technical defaults.

What developers should watch for next

For teams already using uv, Ruff, or ty in production, the near-term advice is simple: nothing changes yet. The deal has not closed, the tools remain open source and under active development, and Astral has continued shipping releases through the announcement period rather than pausing. The things worth watching are the deal’s actual closing date, any governance announcement about how community contributions will be handled afterward, and whether Codex-specific features start appearing in these tools that are not useful outside of OpenAI’s ecosystem.

For engineering leaders evaluating which AI coding agent to standardize on, the deal is a useful data point but not yet a reason to switch. Codex’s bet on owning its Python tooling layer is a multi-year strategy, not something that changes a team’s day-to-day experience this quarter. The more immediate decision factors remain the same ones that mattered before the acquisition: model quality for the languages a team actually uses, pricing that fits usage patterns, and how well an agent integrates with existing CI and review workflows.

Frequently asked questions

What is Astral, and why did OpenAI want to buy it?
Astral is the company behind uv, Ruff, and ty, three widely used open-source Python developer tools covering package management, linting and formatting, and type checking. OpenAI wants to use that team and tooling to make its Codex coding agent more capable across planning, testing, and maintenance, not just code generation.

How much is OpenAI paying for Astral?
Neither company has officially disclosed financial terms. Secondary reporting has put the figure at roughly 750 million dollars, but that number should be treated as an estimate rather than a confirmed price.

Has the deal officially closed?
No. As of October 21, 2026, OpenAI’s own materials describe the transaction as pending, subject to regulatory approval and other customary closing conditions. No specific closing date has been announced.

Will uv, Ruff, and ty stop being open source?
Both OpenAI and Astral have said the tools will remain open source after the deal closes. Astral has continued shipping new releases of all three tools throughout the pending acquisition period.

Who leads Astral and Codex after the deal?
Charlie Marsh is Astral’s founder and CEO. Thibault Sottiaux leads Codex at OpenAI. Astral’s team is expected to join OpenAI’s Codex organization once the transaction closes.

Is this acquisition aimed at Anthropic?
Coverage of the deal has widely framed it as a competitive move against Anthropic, whose Claude Code product has built a strong following among Python developers. Owning Astral gives OpenAI more direct control over a layer of the Python toolchain that Anthropic currently relies on third-party and open tools for.

Does this change how much Codex costs?
Not directly. Codex pricing remains usage-based, with OpenAI estimating average costs of roughly 100 to 200 dollars per developer per month depending on model choice and usage patterns. The Astral deal is aimed at making Codex more efficient and capable rather than changing its pricing structure.

Should developers switch tools because of this acquisition?
Not yet. uv, Ruff, and ty continue to function exactly as before, remain open source, and are still under active development. The acquisition is worth monitoring for governance changes once it closes, but it is not a reason to migrate away from these tools today.

OpenAI’s $750M Astral Deal Still Awaits Approval [2026]