In a major infrastructure move designed to shift artificial intelligence from static text generation to fully autonomous, multi-day automation, OpenAI has announced an agreement to acquire Ona (formerly known as Gitpod GmbH).

The acquisition focuses entirely on the architecture of AI orchestration. By absorbing Ona’s secure, customer-controlled cloud execution technology, OpenAI plans to give its flagship coding assistant, Codex, a permanent, persistent place to work. The move removes the constraint of running AI agents on local user hardware, allowing digital workers to safely execute complex, long-running production software lifecycles in the background.

While the exact financial terms of the deal remain undisclosed, the transaction marks OpenAI’s latest aggressive expansion into core enterprise infrastructure as it positions itself for a highly anticipated initial public offering (IPO).

Moving Past the Laptop: Why OpenAI Needs Persistent Sandboxes

The acquisition addresses a critical structural bottleneck in the current generation of enterprise AI tools: the limitation of local, session-based execution.

Currently, when a software developer or system administrator deploys an AI agent to write code, refactor a repository, or run vulnerability scans, the execution loop is typically tied to their local machine or a temporary browser session. The moment the user closes their laptop, goes offline, or shuts down their workstation, the agent’s environment collapses, terminating any incomplete background processes.

Ona solves this problem by providing cloud-based, isolated sandboxes that remain continuously online.

By shifting Codex operations into Ona’s persistent environments, OpenAI can allow agents to independently execute ambitious, multi-day tasks—such as modernizing massive enterprise codebases, addressing structural security flaws, and executing complex software test suites—entirely in the background. Users can simply close their devices, remain offline, and securely check progress, provide direction, or review final results from a smartphone or tablet days later.

Enterprise Security: Isolating the AI Sandbox

Giving autonomous agents direct access to an enterprise’s code repository presents severe corporate security, data privacy, and compliance risks. If an agent is compromised or hallucinates a command, it could theoretically execute malicious scripts, access sensitive encryption keys, or leak proprietary data to external servers.

Ona’s platform brings a highly robust, customer-controlled security model directly into the OpenAI ecosystem:

• Cryptographic Hashing Guardrails: To block malicious applications from executing within the agent workspace, Ona generates a unique cryptographic signature (hash) for forbidden programs. The platform can instantly detect and block a restricted file even if a bad actor attempts to rename, move, or hide the script.
• Network and File System Isolation: The software enforces rigid boundary controls, preventing AI agents from accessing deep directory paths containing sensitive credentials or making unauthorized outbound connections to external, potentially compromised servers.
• Customer-Controlled Infrastructure: Crucially for enterprise compliance, Ona allows these sandboxes to run natively inside an organization’s own cloud environment. OpenAI provides the model intelligence and core orchestration layer, but the client retains absolute sovereignty over data boundaries, infrastructure logging, and credential scoping.

SEO & Market Dynamics: The Scaling Codex Phenomenon

The acquisition arrives amid staggering growth for OpenAI’s developer ecosystem. The company revealed that Codex now commands over 5 million weekly active users, representing an explosive 400% surge since the start of the year.

Interestingly, the user base is no longer exclusive to software engineers. Non-developers—including financial analysts, operations managers, and marketers—now make up roughly 20% of the total Codex user base, using the engine to execute complex data auditing, corporate finance modeling, and workflow automation.

The deal also marks a fierce counter-response to Anthropic, OpenAI’s closest rival. Anthropic has recently enjoyed massive enterprise momentum behind Claude Code, its highly popular developer tool, pushing Anthropic’s valuation to a massive $965 billion. By integrating Ona’s ready-made infrastructure—which has already supported over 2 million software developers with cloud workspaces—OpenAI can rapidly accelerate the production readiness of its own agent fleet without spending years building custom cloud orchestration layers from scratch.

Until the transaction clears customary closing conditions and required regulatory approvals, both entities will continue to operate as independent companies. Upon official closing, the Ona team will fold directly into OpenAI’s Codex division, completing a critical evolution from conversational chatbots to secure, enterprise-grade digital workforces.

OpenAI Enterprise Architecture: Scaling Agentic Workflows

This enterprise development briefing maps out the structural shift from basic conversational prompting to secure, cloud-hosted developer environments, showcasing how tech companies manage isolation, context, and data governance when deploying autonomous agents.