OpenAI announced on June 11 that it will acquire Ona, a cloud development environment company, to expand Codex with persistent, customer-controlled infrastructure for long-running AI agent workflows. The Ona team will join OpenAI’s Codex organization after the deal closes, according to OpenAI’s announcement.
More than 5 million people now use Codex weekly, up 400% from earlier in 2026, according to OpenAI. The product started as a developer tool and has expanded into research, analysis, and general automation. As agent tasks grow longer and more complex, OpenAI needs infrastructure that lets agents continue working after a user closes their laptop. That is what Ona provides.
What Ona Brings
Ona built cloud development environments for 2 million developers across enterprises including, according to CEO Johannes Landgraf, “the oldest bank in the US, one of Europe’s largest pharma companies, one of Asia’s largest sovereign wealth funds.” The company’s weekly agent sessions grew 13x in production since the start of 2026, Landgraf wrote in a blog post announcing the deal.
The platform provides reproducible cloud environments, scoped credentials, audit trails, agent orchestration, and what the company calls “runtime AI security” through its Veto product, which enforces kernel-level policy controls on agent behavior. Ona deploys inside the customer’s own cloud, giving organizations control over where agents run, what they can access, and how activity is logged.
“Agents need more than intelligence; they need a trusted workspace,” Landgraf said in OpenAI’s announcement. “Joining OpenAI lets us bring that foundation into Codex, helping organizations deploy agents with confidence.”
The Codex Gap
Codex today runs agents tied to a single device or active session. OpenAI’s stated goal with the acquisition is to let agents work over hours or days rather than minutes, continuing progress inside secure cloud environments even when the user who started the task has moved to a different device.
This is the same problem that Microsoft addressed at Build 2026 with its IQ stack and Agent 365 platform, and that Google is tackling through its Agent Skills repository and Managed Agents offering. The pattern across all three companies: frontier model capability alone is insufficient for enterprise adoption. Organizations need infrastructure that handles persistence, governance, credential scoping, and auditability.
OpenAI is choosing to buy rather than build. The Bloomberg report on the deal noted that the acquisition strengthens OpenAI’s ability to support agents at scale, addressing infrastructure and reliability challenges as the company moves deeper into enterprise territory.
Vertical Integration Accelerates
The acquisition follows OpenAI’s Oracle Cloud partnership announced June 10, its confidential S-1 filing on June 8, and its reported negotiations for a 10-gigawatt Ohio data center campus. The company is building a full stack: models, agent products, cloud infrastructure, and compute capacity.
For the enterprise agent market, the signal is clear. Persistent agent execution requires specialized infrastructure that generic cloud providers do not offer out of the box. Ona, Anthropic’s Managed Agents, Microsoft’s IQ stack, and Google’s Agent Skills all converge on the same thesis: agents need governed, persistent, auditable environments to move from demos to production.
The transaction is subject to customary closing conditions, including regulatory approvals. Until close, OpenAI and Ona remain separate companies.