NVIDIA rolls out Codex to ~10,000 staff for internal coding use (GPT‑5.5‑powered)

NVIDIA said on April 23 that more than 10,000 employees are using OpenAI’s Codex internally, powered by GPT-5.5 on NVIDIA infrastructure. (nvidia.com) The users are not just software engineers. NVIDIA said Codex is being used across product, legal, marketing, finance, sales, human resources, operations, and developer programs. (nvidia.com) (qoo10.co.id) Codex is OpenAI’s coding agent, a tool that can read codebases, write code, review pull requests, and work across developer tools. OpenAI says it is built for tasks like feature development, refactors, reviews, and releases. (openai.com) (developers.openai.com) NVIDIA said the GPT-5.5 version of Codex runs on its GB200 NVL72 rack-scale systems. The company said that setup delivers 35 times lower cost per million tokens and 50 times higher output throughput. (nvidia.com) (techspot.com) The company framed the deployment as a shift from coding assistants for individual developers to AI agents used across whole business workflows. NVIDIA executive Justin Boitano wrote that the “next frontier” is knowledge work beyond software engineering. (nvidia.com) OpenAI is making the same pitch to large companies. Last week, it said Codex had reached 4 million weekly active users and announced Codex Labs plus deployment partners including Accenture, Capgemini, Cognizant, Infosys, PwC, and Tata Consultancy Services. (openai.com) That scale changes the problem from model access to workplace controls. Large deployments need repository permissions, audit trails, usage limits, and human review because the agent can read internal code and act across connected tools. (developer.nvidia.com) (openai.com) NVIDIA’s own security team has been warning about that new attack surface. In posts over the past week and past six months, NVIDIA researchers described prompt-injection and dependency-chain risks in agentic coding environments, including attacks that can manipulate instruction files. (developer.nvidia.com 1) (developer.nvidia.com 2) The immediate takeaway is less about one chatbot than about enterprise plumbing. NVIDIA is treating Codex as internal infrastructure, with thousands of employees using an OpenAI agent on NVIDIA hardware inside day-to-day work. (nvidia.com)