The product underneath the libraries

The larger asset is Code Storage. Pierre describes it as a distributed storage system built on Git, with an API designed for machines rather than for a human opening a repository page. On its own product page, Pierre lists repositories up to 32TB, more than 30 concurrent writes, Git LFS support, SDKs in TypeScript, Python and Go, webhooks, and sync with GitHub, GitLab and Bitbucket. Its named customers are mostly AI coding platforms — Lovable, Bolt, Amp, Poke, Anything and Parahelp — storing generated code alongside agent memory files, design artifacts and other files that change over time. Pricing on the same page is usage-based: hot storage at $0.005 per GB per hour, cold storage at $0.0002, with bandwidth charged per GB. Those are Pierre's published figures, not independent measurements.

That customer list is the tell. Code Storage already serves AI product companies building around generated code, agent memory and other changing artifacts. Bringing it in-house gives OpenAI an existing storage product with outside customers, not just another internal infrastructure project.

Two Git problems, two different answers

OpenAI linked the deal to work it described two days earlier. On November 2 it published Pando, the in-house Git backend that now powers its own monorepo, built after parallel work by engineers and Codex agents pushed the old system to slow pushes, failed merges and reliability that at times fell below 90%. Pando is internal and is not a product. Code Storage is the opposite shape: an external product, sold by the hour, designed for other platforms to embed.

OpenAI's announcement does not say the two systems will be combined, and readers should not assume they will be. What the post does state is an ambition to "set the standard for modern software development: from the first branch an agent creates to that magic moment when code is deployed to the world," and, once the deal closes, to "advance the future of source control across the industry." Axios framed the acquisition as a challenge to Microsoft's GitHub. OpenAI's own announcement does not mention GitHub.

What changes, and when

For Codex users, nothing changes on announcement day. The libraries they already use stay in place, OpenAI says Pierre's open-source projects will remain open source and that it will keep contributing to them, and Code Storage customers can keep using the service after closing while its features and functionality are brought into Codex. No timeline is given for that integration, for a closing date, or for what a Codex storage tier would cost.

The practical question is the one the announcement leaves open. Git hosting is where teams keep permissions, history and review habits, and it is sticky in a way a chat window is not. If Code Storage's features arrive inside Codex in a form that outside platforms can no longer license on the same terms, the deal removes a supplier from the market at the same time as it improves Codex. OpenAI's continuity promise covers existing customers after closing. It does not say whether post-acquisition pricing, licensing terms or availability for new customers will change.

Sources

  1. [1] OpenAI, "OpenAI to acquire Pierre Computer Company, Inc.", November 4, 2026 (read first-hand at Gate 1; Trees/Diffs in the Codex app, Highlights, open-source and continuity commitments, closing conditions, no price stated)Read source
  2. [2] Pierre Computer / Code Storage, product page (read first-hand at Gate 1; 32TB repository limit, 30+ concurrent writes, Git LFS, TypeScript/Python/Go SDKs, webhooks, GitHub/GitLab/Bitbucket sync, named customers, 99.99% SLA claim, hot/cold storage and bandwidth pricing — Pierre's published figures, not independent measurements)Read source
  3. [3] OpenAI, "Scaling Git at OpenAI" (Pando), November 2, 2026 (read first-hand at Gate 1; internal monorepo backend, reliability at times below 90%, no mention of Pierre or Code Storage); ALFW coverage: https://alfw.ai/article/openai-pando-git-backend-agents/Read source
  4. [4] Axios, reporting on the acquisition, November 4, 2026, via secondary relays read this run (terms not disclosed; the GitHub-rival framing is Axios's, not OpenAI's)