The challenger accelerates with open-standard Instinct MI350 Series and US $ 6bn OpenAI partnership challenging Nvidia’ s dominance

AMD

The challenger accelerates with open-standard Instinct MI350 Series and US $ 6bn OpenAI partnership challenging Nvidia’ s dominance

AMD emerged as a formidable AI infrastructure challenger in 2025, announcing a landmark 6-gigawatt partnership with OpenAI in October 2025, with first deployments of its MI450 GPUs scheduled for late 2026. The company’ s Instinct MI350 Series, launched at the June 2025 Advancing AI event, delivers a 4x generational AI compute increase and 35x inferencing improvement over previous generation MI300X, while offering 40 % better tokens-per-dollar performance than competing solutions. AMD’ s strategic differentiation centres on open standards and rack-scale architecture, with the MI355X featuring 288GB HBM3E memory supporting models up to 520 billion parameters on a single GPU. Oracle Cloud Infrastructure became the first hyperscaler to deploy MI355X GPUs at scale, while the company’ s ROCm 7 open-source software stack gained significant traction with enterprise customers. AMD’ s preview of the Helios rack-scale platform, built on Meta’ s Open Compute Project design, promises 10x generational performance increases for frontier models with next-generation MI400 Series GPUs, 6th Gen EPYC CPUs and Pensando networking, targeting a 2026 launch.

“At AMD, we’ re really focused on pushing the boundaries of high performance and adaptive computing to help solve some of the world’ s most important challenges”