OpenAI Agrees to Buy Pierre and Bring Its Git-Based Code Storage Into Codex
The developer-tools start-up already powers file trees and code diffs inside Codex, and the deal moves OpenAI further into code hosting, where Microsoft's GitHub dominates.
By Nakoda News Desk
·2 min read
Prefer Nakoda News on Google
OpenAI has agreed to acquire Pierre Computer Company, a developer-tools start-up whose software already runs inside Codex. OpenAI plans to fold Pierre's main product, a Git-based repository service called Code Storage, into its coding agent once the deal closes.
OpenAI announced the agreement on November 4 and did not disclose financial terms. According to reports of the announcement, the deal needs customary regulatory approvals and other closing conditions, and the two companies will run independently until then. Existing Code Storage customers will be able to keep using the service after closing, and Pierre's open-source projects will stay open source.
The deal matters because it takes OpenAI past generating code and into storing and managing it. That is the job of GitHub, the Microsoft-owned platform that has long been where most software teams keep and coordinate their code. Microsoft is also OpenAI's largest backer. Coverage of the deal has cast it as OpenAI joining the companies challenging GitHub, although OpenAI has not described Code Storage as a GitHub replacement and has not said which features will reach Codex users, or when.
What Pierre Builds
Pierre describes itself as a builder of "low-level, composable infrastructure for machine work". According to the company's published material, as summarised by Unite.AI, Code Storage is a distributed storage system built on Git with APIs designed for software agents rather than people. It can hold generated code, agent memory files, design files and other documents. The service supports the standard Git clone, push and fetch commands, offers SDKs for TypeScript, Python and Go, and syncs with GitHub, GitLab and Bitbucket. Pierre says it can handle more than 30 concurrent writes and repositories of up to 32TB.
The named customers include AI app builders Lovable and Bolt, as well as Poke, Anything, Amp and Parahelp. Pierre also maintains the open-source libraries Trees and Diffs, which Codex already uses to display file trees and code changes.
According to a jobs listing for the company, Pierre was founded by Jacob Thornton, a co-creator of the Bootstrap web framework, and Ian Ownbey, an early Twitter engineer. A third-party data aggregator puts its total funding at $23.5 million. Neither company has confirmed that figure.
Agents Strain Version Control
The acquisition follows an OpenAI engineering post on November 2 about Pando, an in-house Git backend it built after Codex sharply increased the number of branches, pushes and merges its engineers were producing. OpenAI said median push latency fell by nearly 80% after the switch and that the system now handles close to a petabyte of Git data a day. It said Pando went from kickoff to production in 10 weeks, was built with Codex, and began serving production traffic in April 2026. OpenAI added that it is studying cases where many agents share one repository and exploring ways to offer the technology to users.
Pierre is OpenAI's latest purchase aimed at Codex. In March it agreed to buy Astral, maker of the Python tools uv and Ruff, and in June it agreed to acquire Ona, which builds the environments that AI agents run in. Both deals were presented as ways to stretch Codex across the full development workflow rather than just writing code.
OpenAI has not given a closing date for the Pierre deal or a timeline for bringing Code Storage into Codex.
Sources
- Unite.AI · OpenAI to Acquire Pierre Computer and Bring Code Storage to Codex
- Investing.com · OpenAI to acquire Pierre to expand Codex developer infrastructure
Produced with AI assistance from the sources above. How we work
By
Nakoda News Desk
Nakoda News covers AI and business. Every story links the sources it draws on; how we work is set out in our editorial standards.
