OpenAI Buys Ona to Let Codex Agents Run for Hours in the Cloud

OpenAI announced on June 11, 2026 that it will acquire Ona, the German cloud company formerly known as Gitpod, to give its Codex agents persistent cloud environments.

According to TechTimes, the deal lets Codex keep working on a task for hours or even days, unattended, after the developer has closed their laptop.

The Problem the Deal Solves

The constraint is architectural. TechTimes reported that Ona provides cloud-based sandboxes, isolated environments where code runs safely, and the key property is that those sandboxes stay online even when the developer who started them shuts down their own machine.

That shifts what an agent can do. Instead of supervising a tool keystroke by keystroke, a developer can hand Codex a job and check on it later, with the work continuing in the background.

The technical fit is direct. Sentinel reported Ona's environments persist across tool calls and sessions, something current stateless execution cannot reliably provide at enterprise scale.

From Gitpod to Ona

The company has a long history. Sentinel reported Ona began life as Gitpod in 2014 in Kiel, Germany, pioneering browser-based remote development environments for human engineers.

Then it pivoted. Sentinel reported the company rebranded to Ona after shifting from remote dev environments for humans to secure cloud execution and orchestration for AI agents, with productive usage growing 13 times over the prior year.

Its customer base signals maturity. Sentinel reported clients span large US banks, European pharmaceutical companies, and Asian sovereign wealth funds, a profile that shows the technology has moved into regulated, risk-sensitive industries.

Codex Is Scaling Fast

The timing tracks Codex growth. CNBC, via MLQ reported Codex now counts more than 5 million weekly active users, up from 3 million in April, a more than sixfold rise since its February desktop launch.

The user base is broadening too. CNBC, via MLQ reported knowledge workers now make up roughly one-fifth of Codex users and are growing three times faster than developers.

That scale creates sustained demand for execution infrastructure that can handle thousands of simultaneous long-running sessions without weakening isolation between enterprise tenants, a pressure tied to the wider economics of AI compute.

The Build-Versus-Buy Logic

Analysts read it as a classic decision. InfoWorld reported one analyst framing it as OpenAI's buy-or-build challenge, where OpenAI has a substantial Codex effort but lacked a safe area to protect enterprise autonomous agent work.

The security plumbing is the hard part. InfoWorld reported Ona delivers a workspace an enterprise can run in its own virtual private cloud, with governance, credential management, and resource access controls that prevent the model from doing what it should not.

There is a competitive read as well. InfoWorld reported one Gartner take describing the deal as OpenAI's response to Anthropic supporting self-hosted sandboxes, part of a broader contest playing out as AI transformation meets its hardest test in security.

What Comes Next

The team moves with the deal. CNBC, via MLQ reported Ona's entire team, led by CEO Johannes Landgraf, will join OpenAI's Codex group once the transaction closes, subject to regulatory approvals.

For operators, the practical shift is toward server-side persistence. Tasks that once died with a closed laptop can run to completion, which changes how teams scope work like testing, debugging, and application modernization.

The close is still pending. Until then the companies remain separate, but the direction is clear, namely that the next phase of AI coding is less about smarter autocomplete and more about agents that keep working while you are away.