AMD formally introduced EPYC 9006 Venice at Advancing AI 2026, saying the Zen 6C server chips will scale to 256 cores and 512 threads per socket.
Venice adds PCIe Gen 6, up to 16 memory channels, DDR5-8000 and Gen2 MRDIMM-12800 support—closing a key Intel Xeon 6 advantage in memory bandwidth.
AMD claimed roughly 20% higher performance per core versus EPYC 9005 Turin and about 1.2x generation-on-generation gains in workloads such as Nginx HTTPS serving.
The lineup includes SP7, value-focused SP8 and Venice-X variants, with top clocks reaching 5.0GHz on a 96-core model; full SKU details were not disclosed.
EPYC 9006 SP7 will not ship until Q4, making the announcement a soft launch as AMD seeks to extend Turin's lead before Intel's Diamond Rapids arrives in 2027.
Can AMD's open software strategy finally break Nvidia's decade-long dominance in the AI industry?
As AI bills soar, is AMD's 'lowest cost per token' claim a real advantage or a pricing illusion?
AMD Helios Debuts with Microsoft Backing: Rack-Scale AI System Targets Nvidia’s Dominance in $75B Data Center Market
Overview
On July 20, 2026, AMD launched its Helios rack-scale AI system, marking a pivotal moment in the AI and HPC industry. Immediately after, Microsoft announced it would expand its Azure AI and HPC infrastructure using AMD technology, highlighting the significance of this launch. Helios extends AMD’s leadership in open hardware and software to the rack level, combining silicon innovation with open, industry-driven design. This strategic advancement positions AMD as a strong competitor in the rapidly evolving AI landscape, offering comprehensive solutions for demanding computational tasks and solidifying its role among top technology providers.