OpenAI and Dell bring Codex to on-premises enterprise environments for the first time
OpenAI and Dell have partnered to deploy the Codex AI coding agent inside hybrid and on-prem enterprise infrastructure, a first for OpenAI.
OpenAI and Dell Technologies announced a partnership on 18 May 2026 at Dell Technologies World to bring Codex into hybrid and on-premises enterprise environments. If you work in a large organisation that has been watching AI coding tools from the sidelines because your data cannot leave your own infrastructure, this one is worth paying attention to.
What is Codex, quickly?
Codex is OpenAI’s software engineering agent. It can write features, fix bugs, answer questions about your codebase, and propose pull requests for review. It runs on codex-1, a version of OpenAI’s o3 model tuned specifically for software engineering tasks. More than 4 million developers now use it every week, and adoption has been expanding beyond pure coding work into things like gathering context across tools, routing product feedback, and preparing reports.
Until now, Codex has been a cloud-first product. This partnership changes that.
What the Dell partnership actually does
The collaboration has two main components.
First, Codex will connect with the Dell AI Data Platform, which many large organisations already use to store, organise, and govern enterprise data on-premises. The integration is designed to give Codex access to internal codebases, documentation, operational knowledge, business systems, and team workflows, without that data needing to travel to a public cloud.
Second, Dell and OpenAI are exploring how Codex, ChatGPT Enterprise, and other API-based solutions can interface with the Dell AI Factory. The work here covers data preparation, systems-of-record management, testing, and deployment of AI applications within hybrid and on-premises Dell infrastructure. Dell already has more than 5,000 customers running the AI Factory, so the distribution reach here is substantial.
Dell Deskside Agentic AI Solutions are already available. Dell AI Data Platform orchestration and search upgrades are scheduled for Q2 2026.
Why this matters: the compliance wall
For most large companies in banking, healthcare, defence, or government, sending internal data to a third-party cloud service triggers compliance reviews, legal sign-offs, data residency checks, and sometimes outright regulatory restrictions. That process can take months, and in some cases, the answer is simply no.
That compliance wall is why a lot of regulated organisations have been sitting out cloud-first AI agent deployments. It is not that they do not want the technology. It is that the data governance requirements make the standard cloud path impractical.
This partnership directly targets that gap. Financial services firms, NHS organisations, and government agencies that have largely been blocked from broader Codex deployment by the on-premises requirement now have a vendor-supported path to running Codex against internal data without exporting it.
What this means for OpenAI’s strategy
This is the first time OpenAI has named a non-Azure, non-cloud distribution path for Codex. Microsoft remains OpenAI’s primary cloud partner through Azure, and that is not changing. But this deal signals that OpenAI is no longer willing to leave the regulated-enterprise segment entirely to Azure’s on-premises and hybrid offerings like Arc.
OpenAI is also working with global systems integrators including Accenture, Capgemini, Cognizant, Infosys, PwC, and Tata Consultancy Services to scale enterprise Codex adoption globally. The combination of a hardware infrastructure partner like Dell and major SI relationships suggests OpenAI is building the enterprise distribution machinery that cloud-only access does not provide.
For Dell, the partnership strengthens its position as enterprises look for places to run agentic AI workloads. AI infrastructure is increasingly where Dell competes, and anchoring a flagship OpenAI product to Dell hardware is a meaningful proof point for that strategy.
The technical picture
Enterprise deployments through this partnership will include standard security controls: SCIM, EKM, role-based access control, domain verification, audit logs, and usage monitoring. These are the controls large IT teams require before they will approve any enterprise software rollout, let alone an AI agent touching internal code and business data.
One area worth watching: Codex’s existing sandbox modes (read-only, workspace-write, and full-access) were designed for code-execution contexts. As Codex expands into broader business workflows touching CRM data, financial records, and operational systems, the permission model will need to evolve beyond that original framing. That is not a criticism, it is just the honest reality of taking a developer tool and extending it into the wider enterprise.
What does this mean for you?
If you work in a regulated industry and your team has been told that cloud-based AI coding tools are off the table for data governance reasons, this partnership gives you something concrete to bring back to your security and compliance teams. The path to running Codex inside your own infrastructure now has vendor backing and a named hardware partner.
If you are an engineering leader or CTO at a large organisation, this is worth adding to your evaluation list for H2 2026. The Dell AI Data Platform integration means Codex could connect with data your teams already manage, rather than requiring a separate cloud migration before you can get value.
If you are already a Dell AI Factory customer, this partnership is directly relevant. The integrations being built are designed for your existing infrastructure.
For everyone else watching from the sidelines: the broader point here is that enterprise AI agent deployment is moving toward the infrastructure organisations already control, rather than requiring organisations to move toward the cloud. That shift is going to change which AI products get adopted in regulated sectors, and which ones stay on the waitlist indefinitely.