AMD EPYC CPUs Power Google Cloud’s Confidential TPU 8tx Infrastructure
5th Gen AMD EPYC processors are powering the host infrastructure for Google Cloud TPU 8tx, which AMD describes as the first cloud Tensor Processing Unit system with hardware-enforced confidential computing. The companies say the combination is designed to help enterprises protect sensitive AI models and data while running workloads in the cloud.
- 5th Gen AMD EPYC processors host Google Cloud TPU 8tx.
- TPU 8tx is described as the first cloud TPU system with hardware-enforced confidential computing.
- AMD SEV helps protect data between the EPYC host and TPU accelerators.
- The system targets inference, fine-tuning, reinforcement learning and agentic AI workloads.
- AMD says EPYC is used by more than 10 leading hyperscalers and frontier AI labs.
EPYC provides the host foundation for TPU 8tx
AMD said on November 09, 2026 that its 5th Gen AMD EPYC processors manage workload orchestration, data preparation and high-throughput communication with the TPU accelerators in Google Cloud TPU 8tx. The expanded collaboration combines EPYC CPU performance, AMD Infinity Guard security capabilities and x86 software compatibility with Google Cloud TPU AI acceleration.
According to AMD, the system is intended to support inference, fine-tuning and reinforcement learning workloads. Google Cloud’s TPU 8tx is part of its AI Hypercomputer, and the companies say the architecture is designed for organizations that need to run sensitive AI workloads at scale while maintaining data protection.
Hardware protections extend between the CPU and accelerators
TPU 8tx uses AMD Secure Encrypted Virtualization, or SEV, capabilities in the EPYC CPU. AMD says these capabilities, working with the TPU, extend confidential computing protection beyond the CPU to connected accelerators.
The company says customer prompts, model weights and intermediate data can remain encrypted in CPU memory and while moving between the host and accelerators. Hardware isolation is used during accelerator execution, while an accelerator’s identity and security configuration are verified before it receives direct access to protected memory.
These measures are aimed at reasoning and agentic AI applications that process sensitive information. AMD also said its EPYC processors support the broader Google Cloud confidential computing portfolio, including Confidential VMs, Confidential GKE, Confidential Space, Confidential Managed Service for Apache Spark and Confidential Dataflow.
AMD positions EPYC as a cross-accelerator platform
AMD said EPYC processors are the host CPU of choice across more than 10 leading hyperscalers and frontier AI labs. The processors support systems built with AMD Instinct GPUs, Nvidia GPUs and custom AI accelerator ASICs.
In AMD’s description, host CPUs remain responsible for supplying data, scheduling work and connecting accelerated AI workloads with surrounding applications and services. The company is presenting EPYC’s x86 software compatibility and use across different accelerator architectures as a way to give customers a consistent CPU platform as AI infrastructure becomes more specialized.
The TPU 8tx deployment builds on multiple generations of AMD and Google Cloud collaboration. The companies have worked together on general-purpose, compute-intensive, storage-intensive and accelerated workloads, including Google Cloud Compute Engine C4D, N4D, H4D, Z4D and G4 virtual machines. For users, the announcement means Google Cloud’s TPU acceleration is being paired with a hardware-based confidentiality model intended to protect data while it is being processed and exchanged with the host system.
