OpenAI's GPT-5.5 Powers Codex on NVIDIA Systems, Transforming Workflows
OpenAI has unveiled GPT-5.5, its latest AI model, which powers the Codex application and runs exclusively on NVIDIA’s GB200 NVL72 infrastructure. Over 10,000 NVIDIA employees, spanning roles from engineering to marketing, have already integrated GPT-5.5 into their workflows, reporting transformative productivity gains.
Codex, OpenAI’s agentic coding platform, leverages GPT-5.5 to streamline complex tasks. According to NVIDIA, debugging cycles that used to take days now take mere hours, while feature development from natural-language prompts has become significantly faster and more reliable. The GB200 NVL72 systems supporting this deployment deliver 35x lower cost per million tokens and 50x higher token throughput per second per megawatt compared to previous-generation systems, making enterprise-scale AI inference economically viable.
Jensen Huang, NVIDIA’s founder and CEO, highlighted the importance of this partnership in a company-wide email, urging employees to embrace the Codex app. "Let’s jump to lightspeed. Welcome to the age of AI," he wrote.
Enterprise Security at Scale
For enterprises considering AI adoption, security is a critical concern. NVIDIA’s deployment addresses this with robust safeguards, including dedicated virtual machines for each user. These cloud-based sandboxes ensure that agents operate securely while maintaining full auditability. A zero-data retention policy and read-only access to production systems further protect sensitive information.
This secure, scalable setup has made it easier for NVIDIA to integrate Codex across departments without compromising data integrity or operational workflows, setting a benchmark for enterprise AI adoption.
A Longstanding NVIDIA-OpenAI Partnership
The release of GPT-5.5 is the culmination of a decade-long collaboration between OpenAI and NVIDIA. Their partnership began in 2016, when Huang personally delivered the first NVIDIA DGX-1 AI supercomputer to OpenAI. The two companies have since worked closely on hardware-software co-design, with OpenAI providing feedback that has shaped NVIDIA’s product roadmap.
Notably, OpenAI is committing to deploy over 10 gigawatts of NVIDIA systems to support its next-generation AI infrastructure. This ambitious buildout will involve millions of GPUs and positions NVIDIA as a foundational partner for OpenAI’s future advancements in model training and inference.
One tangible result of this collaboration is the first GB200 NVL72 100,000-GPU cluster, which has already set new benchmarks for system reliability and scalability at frontier AI levels. This infrastructure has been instrumental in delivering GPT-5.5’s capabilities.
Market Implications
While OpenAI and GPT-5.5 are not publicly traded entities, the partnership with NVIDIA underscores the growing demand for advanced AI infrastructure. NVIDIA’s dominant position in the AI hardware market, supported by partnerships like this, could further enhance its role as a critical enabler of AI-driven innovation. Investors and industry stakeholders should monitor how these developments translate into broader enterprise adoption, as it could signal significant growth opportunities for NVIDIA.
For now, GPT-5.5’s performance within NVIDIA highlights what’s possible when cutting-edge AI models meet optimized hardware. As enterprises race to integrate AI into their operations, this deployment offers a glimpse of what the future of work may look like: faster, smarter, and more secure.