Defining the OpenAI Ona acquisition and the rise of persistent agents
The OpenAI Ona acquisition is a strategic move to bring secure, customer-controlled cloud execution into Codex so AI agents enterprise users rely on can run long-lived, stateful workflows beyond a single device or session. Instead of being limited to brief interactions tied to an open laptop or browser tab, Codex agents will gain persistent environments where they can continue working for hours or days on complex tasks. This shift responds to rapid Codex adoption: OpenAI reports that weekly Codex usage now exceeds five million people, a 400% increase from earlier in 2026, spanning software development, research, analysis and workflow automation. By integrating Ona’s cloud infrastructure, OpenAI aims to close a clear infrastructure gap between powerful language models and the reliable, compliant execution layers enterprises need for production-scale, long-running workflows driven by AI agents.
From 400% Codex growth to demand for long-running workflows
Codex has expanded from an AI coding assistant into a broader platform for research, analysis and workplace automation, and this growth is reshaping expectations for AI agents enterprise teams deploy. Weekly usage now tops five million users, representing a 400% jump from earlier in 2026, and many of those workloads no longer fit into short, interactive sessions. Software testing, issue resolution, vulnerability fixes and application modernization can require hours or days of continuous activity. Existing Codex workflows, however, often depend on a user’s active device, which interrupts or terminates work when sessions end. OpenAI is redesigning Codex around assignments that run from initial request to completed output, even when the work spans multiple human check-ins. The Ona acquisition is the infrastructure response to this pattern: it supplies the persistent cloud execution that Codex’s long-running workflows now require.
How Ona’s cloud execution closes the persistent agent gap
Ona has built cloud infrastructure that moves development and automation tasks off local machines into reproducible, persistent environments, supporting around two million developers so far. For Codex, this means agents can maintain state, keep context and access the right tools over extended periods without being tied to a single device. Sarah Friar, OpenAI’s Chief Financial Officer, described the impact clearly: “Ona will help change that, giving Codex secure, persistent environments where agents can access the tools, systems, and context they need, even when laptops are closed.” In practice, users will be able to start an assignment, close their devices, and later review progress, add instructions or approve decisions while the agent continues working in the background. This architecture turns Codex into a system capable of sustained execution, not only bursty assistance, which is essential for complex enterprise workflows.
Customer-controlled cloud environments for enterprise AI infrastructure
A central promise of the OpenAI Ona acquisition is a customer-controlled execution model that aligns with enterprise AI infrastructure requirements. Instead of running agents in vendor-hosted environments, organizations will host Codex-driven agents inside their own cloud, while OpenAI supplies models and orchestration. Enterprises can decide where agents run, which internal systems they can reach, how credentials are constrained, what activity is logged and when work must go to human review. This addresses governance, access control and compliance concerns that emerge when pilots evolve into production deployments. Johannes Landgraf, Ona’s co-founder and CEO, summed up the design goal: agents need “a trusted workspace” with the context, control and collaboration enterprises require. By combining Codex with Ona’s model, OpenAI is turning AI agents from experimental tools into governed, auditable components of long-running enterprise workflows.
Competing in enterprise AI agents and workflow automation
By adding Ona’s secure, persistent cloud execution to Codex, OpenAI is stepping beyond foundation models into enterprise AI infrastructure and workflow automation. Future Codex deployments will not only generate code or answers; they will execute full software lifecycle tasks, from automated testing to issue resolution and application modernization, over extended periods. Thibault Sottiaux, Core Products Lead at OpenAI, notes that enterprises “want powerful agents that can do real work while meeting the security and control requirements of their environments,” and Ona is central to that promise. The acquisition also signals a competitive shift: AI vendors are building complete operational stacks for AI agents enterprise customers can trust in production. As Ona’s team joins the Codex organization, OpenAI is positioned to offer a tightly integrated stack for Codex cloud execution, persistent agents and governed long-running workflows.
