Oracle to deploy 50,000 AMD AI chips in 2026 supercluster

Oracle (NYSE: ORCL) announced it will deploy 50,000 AMD Instinct MI450 Series GPUs in a publicly available AI supercluster beginning in the third quarter of 2026. The deployment will expand further in 2027 and beyond, according to a press release issued by both companies.

The supercluster will utilize AMD's (NASDAQ: AMD) "Helios" rack design, incorporating AMD Instinct MI450 Series GPUs, next-generation AMD EPYC CPUs codenamed "Venice," and AMD Pensando advanced networking technology codenamed "Vulcano." Oracle stated it will be the first hyperscaler to offer this configuration publicly.

Each AMD Instinct MI450 Series GPU will provide up to 432 GB of HBM4 memory and 20 TB/s of memory bandwidth. The architecture includes liquid-cooled, 72-GPU racks designed for performance density and energy efficiency.

Oracle also announced general availability of OCI Compute with AMD Instinct MI355X GPUs, which will be available in Oracle's supercluster that can scale to 131,072 GPUs.

"Our customers are building some of the world's most ambitious AI applications, and that requires robust, scalable, and high-performance infrastructure," said Mahesh Thiagarajan, executive vice president of Oracle Cloud Infrastructure.

Forrest Norrod, executive vice president at AMD's Data Center Solutions Business Group, stated that the partnership aims to accelerate AI with "open, optimized, and secure systems built for massive AI data centers."

The announcement was made at Oracle AI World in Las Vegas. The companies have collaborated for multiple years, with Oracle previously launching AMD Instinct MI300X powered shapes in 2024.

You May Also Be Interested In