OpenAI’s New GPT-5.5 Powers Codex on NVIDIA Infrastructure — and NVIDIA Is Already Putting It to Work

2026-04-23 · Source: NVIDIA Blog · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering, Cloud Computing & IT Infrastructure · Depth: Intermediate, quick

Summary

NVIDIA has deployed OpenAI's GPT-5.5-powered Codex, an agentic coding application, across its enterprise, with over 10,000 NVIDIANs utilizing it for various functions. Running on NVIDIA GB200 NVL72 rack-scale systems, GPT-5.5 offers significant economic advantages, including 35x lower cost per million tokens and 50x higher token output per second per megawatt compared to previous-generation systems. This deployment has dramatically accelerated developer workflows, reducing debugging cycles from days to hours and enabling rapid experimentation. The Codex app operates within a secure enterprise environment, supporting SSH connections to cloud VMs, ensuring data security with a zero-data retention policy and read-only access to production systems. This initiative highlights a decade-long collaboration between NVIDIA and OpenAI, which began in 2016 with the delivery of the first NVIDIA DGX-1.

Key takeaway

For CTOs and VPs of Engineering evaluating enterprise AI agent deployments, NVIDIA's experience with GPT-5.5-powered Codex demonstrates substantial gains in developer efficiency and cost-effectiveness. You should prioritize secure, auditable deployments using dedicated cloud VMs and consider the economic benefits of advanced hardware like the GB200 NVL72 to make frontier model inference viable at scale, accelerating your teams' innovation cycles.

Key insights

AI agents powered by frontier models significantly enhance enterprise productivity and innovation, especially in coding and knowledge work.

Principles

• Dedicated compute for each AI agent enhances security.
• Full-stack collaboration accelerates AI infrastructure development.

Method

NVIDIA deployed cloud VMs for each employee's Codex agent, ensuring a dedicated sandbox with read-only production access and a zero-data retention policy for maximum security and auditability.

In practice

• Utilize agentic tools for complex, multi-file codebases.
• Implement secure cloud VMs for agent deployment.
• Integrate AI agents for end-to-end feature development.

Topics

• GPT-5.5
• OpenAI Codex
• NVIDIA GB200 NVL72
• AI Agents
• Enterprise Security

Best for: CTO, VP of Engineering/Data, Investor, AI Engineer, AI Architect, Director of AI/ML

Related on AIssential

• Is GPT-5.5 Delivering ‘Life-Changing’ Results for NVIDIA? — GPT-5.5 on NVIDIA GB200 NVL72 systems significantly boosts enterprise productivity and accelerates complex engineering workflows.
• Codex + GPT-5.5 = SUPER APP! Build and Do ANYTHING! — OpenAI's GPT 5.5 and Codex offer a superior, more efficient AI coding and automation workflow.
• OpenAI gpt-oss — OpenAI's gpt-oss models offer powerful local AI with agentic features and efficient quantization under an Apache 2.0 license.
• [AINews] GPT 5.5 and OpenAI Codex Superapp — AI models are advancing in intelligence and cost-efficiency, driving a shift towards agentic workflows and open-source competition.
• OpenAI Drops GPT-5.5 — GPT-5.5 offers faster, more cost-efficient AI processing, advancing OpenAI's "super app" vision.
• OpenAI launches GPT-5.4 Thinking and Pro combining coding, reasoning, and computer use in one model — GPT-5.4 unifies coding, reasoning, and computer operation, significantly boosting performance in professional knowledge work and desktop navigation.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by NVIDIA Blog.