NVIDIA and Google Cloud Expand Collaboration to Advance Agentic and Physical AI

NVIDIA and OpenAI have deepened their decade-long partnership to launch GPT-5.5, OpenAI's latest frontier model, which now powers the Codex agentic coding application. The model runs on NVIDIA GB200 NVL72 rack-scale systems, delivering 35x lower cost per million tokens and 50x higher token output per second per megawatt compared to prior-generation systems. This efficiency makes frontier-model inference viable at enterprise scale, enabling over 10,000 NVIDIA employees across engineering, product, legal, marketing, and other departments to use Codex for tasks like debugging, experimentation, and shipping end-to-end features from natural-language prompts. NVIDIA CEO Jensen Huang urged employees to adopt Codex, stating, "Let's jump to lightspeed. Welcome to the age of AI."

For enterprise security, Codex supports remote SSH connections to approved cloud virtual machines, allowing agents to work with real company data without external exposure. NVIDIA IT rolled out dedicated cloud VMs per employee, providing a sandboxed environment with zero-data retention and read-only permissions via command-line interfaces and Skills. The partnership, which began in 2016 when Huang delivered the first DGX-1 to OpenAI, now includes OpenAI committing to deploy over 10 gigawatts of NVIDIA systems for next-gen AI infrastructure. OpenAI also provided feedback that informed NVIDIA's hardware roadmap, leading to the joint bring-up of the first GB200 NVL72 100,000-GPU cluster, which set a new benchmark for system-level reliability at frontier scale.