OpenAI just bought a cloud execution company called Ona to solve the problem that every Codex user has hit: agents can't run anything that takes more than a few minutes.
Codex Agents Needed a Runtime, Not Just a Sandbox
Codex turns natural language into code, but it has always been a step removed from actually running that code in a persistent environment. OpenAI's own hosted playgrounds and third-party sandboxes cap execution time and prevent agents from maintaining state across long-running workflows. Ona fills that gap—it's a cloud execution firm designed for exactly this: spinning up secure, isolated environments where AI agents can run complex, multi-step tasks without timing out or leaking state.
Ona's Team Brings Production-Grade Cloud Execution
Ona isn't just a product; it's a team that has been shipping infrastructure for running untrusted code safely at scale. Post-acquisition, Ona's engineers join OpenAI to embed that runtime directly into Codex. The source material says the acquisition "aims to support growing Codex usage and help enterprises deploy AI in production." That means OpenAI is betting that the biggest barrier to Codex adoption isn't model quality—it's the lack of a reliable, secure execution backend.
What This Means for Enterprise AI Deployment
Enterprises won't adopt code-generating agents that can't run the code they generate. With Ona, Codex gains the ability to provision cloud VMs, install dependencies, execute long-running scripts, and return results—all inside OpenAI's managed environment. No more copy-pasting to your own AWS account. This is the infrastructure play that turns Codex from a fancy autocomplete into an autonomous systems engineer. Expect Codex agents to get significantly longer timeouts and the ability to leave behind stateful cloud artifacts.
Source: OpenAI to acquire Ona to strengthen Codex cloud capabilities
Domain: economictimes.indiatimes.com
Comments load interactively on the live page.