OpenAI and Dell Technologies announced on May 18, 2026 a new partnership that brings the Codex coding agent to Dell's enterprise infrastructure[1][2]. Under the deal, Codex will integrate with Dell's AI Data Platform and AI Factory so it can run in hybrid and on-premises environments, and the two companies will also explore deeper Codex connections with the Dell AI Factory[1]. The announcement is part of a broader set of ecosystem expansions unveiled at the opening of Dell Technologies World 2026, with most products and integrations rolling out throughout 2026[2].
Codex integrated into Dell's AI Data Platform and AI Factory
According to Reuters, OpenAI and Dell are bringing Codex to "hybrid and on-premise" enterprise environments and will look into how Codex can connect with the Dell AI Factory[1]. SiliconANGLE, reporting from the Dell event, says Codex will integrate with Dell's AI Data Platform and AI Factory infrastructure, giving customers a way to run an AI coding agent inside their own environment without pushing code or data out to a public cloud[2].
Dell says its AI Factory with Nvidia initiative now has more than 5,000 customers globally as of May 2026, providing a sizable enterprise base into which Codex can plug as a coding agent layer for automated code generation and fixes[2].
Pain of "execution" and cost pressure fuel on-prem AI
Dell framed the move around a recurring enterprise pain point: companies are stuck not on AI ideas but on actually executing them at scale. Sam Grocott, senior vice president of product marketing at Dell, told reporters that most enterprises do not have an AI ambition problem but an AI execution problem[2].
Cost concerns came with a concrete example. Jon Siegal, senior vice president of Dell's client solutions group, said a single developer once burned through 1 billion tokens in 24 hours, generating a cloud bill of about $3,400 (roughly 540,000 yen), and used the case to highlight how agentic AI keeps consuming inference tokens continuously[2]. ※ Calculated at 1 USD = 159 JPY[3]
Dell's newly announced Dell Deskside Agentic AI is positioned as the on-prem response: it runs open-weight models locally while keeping data inside the customer's environment, and Dell claims it can cut spend by up to 87% versus public cloud over a two-year period[2]. The high-end workstation configurations are said to support models up to 1 trillion parameters, providing a credible local home for coding agents like Codex[2].
One of several deals alongside Google, Palantir, ServiceNow and Hugging Face
The OpenAI partnership is part of a larger ecosystem package Dell rolled out at Dell Technologies World 2026. Dell says Google's Gemini models will be available through Google Distributed Cloud running on Dell infrastructure, and Palantir's Foundry and Artificial Intelligence Platforms are also moving on-premises onto Dell hardware[2]. Tie-ups with ServiceNow and Hugging Face were announced at the same time, with Codex effectively filling the "coding agent" slot in this lineup of enterprise AI partnerships[2].
A new Dell AI Ecosystem program will validate and certify partner applications running on Dell infrastructure[2]. For OpenAI, this gives Codex a path beyond ChatGPT-hosted cloud usage and into large enterprises through Dell's sales channel and systems-integrator network, a meaningful move in the commercial AI coding race against tools like Microsoft's GitHub Copilot and Anthropic-based coding agents.
Several details remain open. The partnership announcement does not yet specify Codex on Dell's exact availability date, supported model lineup (mix of GPT-5-series, o-series and Codex-specific models), licensing structure, or support scope. Dell only says that most newly announced products and integrations will become available throughout 2026, with some shipping immediately[2].
Summary
The OpenAI–Dell deal pushes Codex from "a cloud agent inside ChatGPT" toward "a developer infrastructure piece that can live inside enterprise on-prem and hybrid environments." For Dell, it puts a strong coding agent at the front of its AI Factory pitch; for OpenAI, it opens a realistic route to reach large companies that, due to data sovereignty or regulatory pressure, are reluctant to go all-in on public-cloud AI. Pricing and support details are still pending, but the announcement noticeably widens the enterprise deployment options for AI coding tools.
Source [3]: https://www.boj.or.jp/en/statistics/market/forex/fxdaily/fxlist/index.htm
