AMD Publishes Official Benchmarks for 256-Core EPYC 9996, ~73% Generational Throughput Gain
Summary
In its September 17 "EPYC Venice Architecture and Performance Overview" whitepaper, AMD published benchmark results for its flagship sixth-generation EPYC 9996 processor. The chip offers 256 cores and 512 threads per socket, targeting cloud computing, virtualization, enterprise databases, HPC and the CPU host side of AI infrastructure workloads. AMD says its SPECrate 2026_int_base overall throughput is 2.37x that of Intel's Xeon 6980P and 2.24x that of Nvidia Vera, and about 73% higher than the previous-generation 192-core flagship EPYC 9965.
Bubble analysis
This is a server-CPU benchmark story with no investment figures, valuations or capex data, so it offers almost no direct evidence on whether AI is in an investment bubble. At most it shows hardware competition in AI infrastructure is still iterating quickly, but faster chips are not the same as returns on investment, and say nothing about whether spending is running ahead of revenue.
Related signals
linked by shared tags
- Nscale Files U.S. IPO Targeting $35B Valuation Despite $1B Loss#datacenter#nvidia
- Interview with Huawei's Wang Tao: Building AI Compute Competitiveness with Super Nodes and Clusters#chips
- A Visit to Xinjiang Karamay Cloud Computing Industrial Park: Feeling the New Momentum of Computing Power Development#datacenter
- China's No.1 Province for Compute: 88 Large-Scale Compute Centers, Intelligent Compute Scale Leads Inner Mongolia and Zhejiang#datacenter
- Tech Giants Call for an AI Slowdown; Acer Chairman Chen Junsheng Says It's Unlikely — Technology Will Only Accelerate#nvidia