OpenAI has announced plans to deploy AMD's new Helios rack-scale system through multiple deployment partners starting in the second half of 2026. This rollout marks the first phase of a massive up to 6 gigawatt (GW) AMD GPU deployment partnership originally announced in October 2025, which also granted OpenAI the option to acquire up to 10 percent of AMD.
The Helios system, officially launched today, combines AMD Instinct MI455X GPUs, 6th Gen AMD Epyc 'Venice' server CPUs, AMD ROCm software, and AMD Pensando networking into a single, integrated rack-scale architecture. OpenAI confirmed it has already had trial access to Helios "for the past three months," following its previous deployments of MI300X and MI355X GPUs.
"We expect to be deploying Helios at massive scale," said Sachin Katti, OpenAI's Vice President of Compute Strategy. The deployment underscores the deepening strategic relationship between the two companies, as OpenAI seeks to secure the massive compute resources required for training and running advanced AI models. The partnership comes amid a broader industry race to secure high-performance AI hardware, with AMD also recently announcing an up to 2GW deal with Anthropic and an investment in that company earlier this week.
OpenAI to Deploy AMD Helios Rack-Scale System in First Phase of Up to 6GW Partnership
By: IDCNOVARegion: North America