AMD to Supply 5th Gen EPYC CPUs for Google's Next-Gen AI System

What happened

On Nov. 9 (local time), AMD announced that it will supply its EPYC server CPUs for Google's next-generation AI computing system. The system is Google's next-generation TPU 8tx, based on the eighth-generation Tensor Processing Unit, combining Google's TPU with AMD's fifth-generation EPYC processor, codenamed Turin, and ultra-high-speed networking. EPYC serves as the host CPU, overseeing overall system operations including AI workload orchestration, data preparation and delivery, and high-speed communications management. The system also incorporates AMD Secure Encrypted Virtualization (SEV), built into EPYC CPUs, extending protection that previously covered only the CPU and its directly connected memory to the connected TPUs: data remains encrypted while stored in CPU memory and during transfers between the CPU and TPU, with hardware isolation preventing external access during computations inside the TPU. AMD states its EPYC processors are used by more than 10 major cloud service providers (hyperscalers) and advanced AI research laboratories, support systems based on AMD Instinct GPUs, Nvidia GPUs and ASIC chips, and are compatible with the existing x86 software ecosystem. AMD and Google have previously collaborated on deploying multiple generations of EPYC processors in Google Cloud and on enterprise security technologies. This is an announcement of a supply arrangement; the source does not state the cause, and no shipment volumes or financial terms are reported.

Impact and signal

Affected components and suppliers

Source and time

Source: The Elec · 11/10/2026 · View original source