AMD EPYC 9006 "Venice-LP" is a purpose-built enterprise processors for rack-scale AI solutions. It is designed with the exact same system philosophy as NVIDIA's Grace and Vera CPUs—to handle serial processing loads and the OS stack for AI GPU servers with bandwidth-heavy network stacks. While NVIDIA uses specialized Arm CPU cores, AMD remains an x86 loyalist, and designed "Venice-LP" around its latest "Zen 6" microarchitecture.

The CPU side of EPYC 9006 LP consists of a 72-core/144-thread "Zen 6" complex that runs at a boost frequency of up to 5.00 GHz. The processor features a 24-channel LPDDR5X memory interface that supports SOCAMM2 modules. The main interconnect between the processor and AI GPUs is 112 Gbps-per-pin xGMI, along with optimization for AMD Instinct MI455X GPUs that are powered by the CDNA 5 architecture.Unlike the rest of the EPYC 9006 family, the 9006 LP isn't a socketed processor, rather, it is an embedded, BGA package, similar to NVIDIA Vera CPU. The chip has a high-degree of power and thermal optimization specific to the platform it's being designed for.

Although announced today, it won't be until the second half of 2027 that the first systems with EPYC 9006 LP processors will arrive. AMD will build compute nodes with Instinct MI455X CDNA 5 AI GPUs and EPYC 9006 LP processors, along with various configurations of LPDDR5X SOCAMM2 memory.