AMD Unveils Helios, Its Next-Gen AI Powerhouse With MI455X & 6th Gen EPYC, Challenging NVIDIA’s Rack-Scale Dominance - Wccftech

What the report says

Wccftech reports that AMD has detailed Helios, a next-generation rack-scale AI platform built around its Instinct MI455X accelerators, 6th Gen EPYC “Venice” processors and Pensando networking components. The system is presented as AMD’s first full-stack AI rack offering, aimed at large-scale inference, frontier AI and sovereign computing customers. The article says AMD has shown the platform publicly ahead of an expected formal launch at its Advancing AI 2026 event.

According to the report, each Helios rack uses 72 MI455X GPUs and 256-core EPYC CPUs, with total HBM4 memory listed at 31 TB. Wccftech says the MI455X is based on AMD’s CDNA 5 architecture and supports high-throughput AI formats, including FP4 and FP8, while moving from HBM3e to HBM4 for higher memory capacity and bandwidth. The publication also cites an estimated rack price of about $5 million to $5.5 million.

The CPU side is described as relying on AMD’s Zen 6-based EPYC Venice chips, which Wccftech says are designed for high-performance computing and AI workloads. The report also points to AMD’s ROCm software stack, Infinity Fabric and Pensando data-processing and networking technology as parts of the broader platform.

The significance is competitive: Wccftech frames Helios as AMD’s bid to challenge Nvidia’s dominance in rack-scale AI systems, including current and future Nvidia rack platforms. More broadly, rack-level designs are becoming increasingly important as AI companies seek integrated systems that combine accelerators, CPUs, networking and software for training and inference at data-center scale.