OpenAI has integrated AMD Epyc Turin generation processors into its server infrastructure to support its proprietary Jalapeno AI chip. Each server node is equipped with 1.5 TB of RAM to handle computational workloads.
Company executives stated that the decision to prioritize AMD hardware over Nvidia Vera was driven by a need for stability and risk mitigation. The Epyc platform was selected for its maturity and the existing expertise of the company's partners.
The development team emphasized that the Jalapeno program focused on pragmatic engineering choices. By opting for established CPU technology, OpenAI aimed to balance performance and cost while avoiding the uncertainties associated with newer, less mature solutions.
Nvidia Vera processors are currently in the early stages of market availability through partners. The assessment from OpenAI represents one of the first public evaluations of the platform by a major technology firm.
