- AMDS MI400 GPU is expected to offer 432 GB HBM4 across 12 memory stacks
- 256-core EPYC Venice CPU for debut with PCIe Gen6 and 2NM process
- AMDS HELIOS RACK aims at 10x performance with double wide AI architecture
AMD recently revealed its instinct MI350 series of GPUs, but the bigger news is what the chip giant has planned by 2026.
The company is preparing for a next generation AI platform that includes the instinct MI400 GPU, EPYC “Venice” CPU and a major shift in rack design with its Helios infrastructure, an expansive, double wide configuration aimed at scaling performance and bandwidth.
The MI400 GPU is expected to be sent by up to 432 GB HBM4 memory, built using 12 stacks with 36 GB HBM4, based on numbers shared by microns and memory per day. GPU estimates from AMD’s rack capacity.
A direct response to nvidia
This would be a significant jump from the current MI350’s 8-Stack setup, which puts it on track for Rival Nvidia’s upcoming Vera Rubin platform.
Alongside the MI400, AMD plans to debut its 256-core EPYC “Venice” server CPU, built on a 2NM process with PCIe Gen6 support and up to 1.6 TB/S memory bandwidth.
The platform will also include Vulcano 800G NIC and support open standards such as Ualink and Ultra Ethernet for improved scale-out connection.
AMD says Helios Rack and the MI450 GPU will offer up to 10x performance of MI355X and place it as a direct response to Nvidia’s accelerated roadmap.
Sam Altman appeared on stage at the company’s recent advancing AI event with AMD’s CEO Lisa SU to emphasize Openai’s early interest in the new platform and said “I think it will be an amazing thing.”
Although hardware first arrives before 2026, AMDS Preview sets the tone of what can be a hard breed with Nvidia.
Helios -Rack may not have an official name yet, but it is clear that AMD is thinking bigger – and wider! – When it moves towards the next general AI infrastructure.
Via Servethehome
