OpenAI to Buy Ona to Power Persistent Codex Agents

OpenAI on June 11, 2026 announced it will acquire Ona, a cloud startup that supplies secure, persistent execution environments for AI agents, folding the company’s runtime into OpenAI’s Codex product line to enable agents that can run across corporate clouds.

Ona — the company that rebranded from Gitpod in 2025 — has repositioned itself from browser‑based development environments to a platform for autonomous software‑engineering agents and isolated workspaces. The move to join OpenAI follows months of rapid enterprise adoption.

Technically, Ona builds pre‑configured, sandboxed environments that include source, dependencies, tools and scoped credentials so code and agents can run securely outside a developer’s laptop. OpenAI said it will integrate that execution layer into Codex so agents can maintain state and keep working when users disconnect.

OpenAI framed the deal as part of Codex’s expansion: the company said more than 5 million people now use Codex weekly, and usage has climbed roughly 400% since early 2026 as the product moves beyond short sessions to tasks that take hours or days.

A key claim from OpenAI and Ona is that customer‑controlled execution lets enterprises run agents inside their own cloud accounts, giving IT teams control over where code executes, which resources are accessible, and how activity is logged and reviewed. That model aims to address common enterprise concerns about data access and governance.

Ona’s leadership framed the tie‑up as a way to scale the platform’s reach. In a company post, CEO Johannes Landgraf said joining Codex will let Ona’s technology compound with OpenAI’s research and distribution to make agentic work safe and enterprise‑ready.

The acquisition underscores a broader engineering challenge: agents that can do useful, long‑horizon work need persistent execution, reliable recovery, and verifiable logs — not just model improvements. Industry analysts have argued that long‑running agents require lifecycle, memory, and interface controls to be production‑grade.

For enterprises, the pitch is practical. Payments and fintech outlets reporting the deal highlighted how persistent agents could automate multi‑step workflows — from incident investigations to nightly test suites — while preserving audit trails and access limits that regulators demand.

OpenAI’s announcement made clear the acquisition is subject to customary closing conditions and regulatory approvals; until the deal closes OpenAI and Ona will operate independently. After closing, Ona’s team will join the Codex organization to continue developing enterprise execution capabilities.

Neither company disclosed financial terms in initial filings and press materials. Reporting outlets and industry briefings say the companies have not published a purchase price. Analysts will watch whether the deal is structured as an asset buy, stock purchase, or strategic hire.

Customers and enterprise buyers who have piloted Ona’s systems told the company that the platform helped move workflows off single machines and into reproducible cloud workspaces. OpenAI and Ona positioned that capability as central to scaling Codex beyond ad hoc coding help to sustained process automation.

What to watch next: regulatory review timelines, the speed of technical integration, and how OpenAI manages tenancy and data‑residency options for large customers. The acquisition signals that the next phase for coding assistants is less about single replies and more about agents that can run real work continuously and under enterprise controls.