Nvidia Deploys OpenAI Codex AI Agent Across Entire Workforce on Blackwell Infrastructure

Gate News message, April 25 — Nvidia has rolled out OpenAI’s Codex, an AI agent powered by GPT-5.5, to its entire workforce following a successful trial with approximately 10,000 employees, according to internal communications from CEO Jensen Huang and OpenAI CEO Sam Altman.

Codex is designed to assist with coding, planning, and workflow tasks across multiple departments including engineering, product, legal, marketing, sales, and operations. The system runs on Nvidia’s Blackwell infrastructure, and the companies have established a Codex Lab dedicated to training and development. OpenAI revealed that Codex has surpassed 4 million weekly active developers.

Nvidia reports significant efficiency gains from running GPT-5.5-powered Codex on GB200 NVL72 rack-scale systems: cost per million tokens decreased by 35x compared to earlier systems, while token output per second per megawatt increased by 50x. For security, Nvidia employs zero-data retention policies and provisions staff with cloud virtual machines (VMs). The Codex application connects via remote Secure Shell (SSH) to approved cloud VMs, with agents receiving read-only access to production systems. This architecture enables Codex to handle tasks running longer than 24 hours by using context compression to maintain memory efficiency.

Beyond Nvidia, Codex is expanding into broader enterprise software adoption. OpenAI is pursuing similar rollouts with other major companies; Canadian technology services firm CGI reports tens of thousands of its engineers, experts, and consultants already using OpenAI tools including Codex. Nvidia and OpenAI are also deepening collaboration on chip design, with OpenAI helping test and refine future architectures while gaining early access to new system designs. This partnership extends a nonbinding letter of intent for OpenAI to deploy at least 10 gigawatts of Nvidia systems for next-generation AI infrastructure.