AMD EPYC Expands Its Leadership at the Foundation of AI

AMD (NASDAQ: AMD) revealed that 5th Gen AMD EPYC processors will serve as the host infrastructure for Google Cloud’s TPU 8tx, the first cloud Tensor Processing Unit system to offer hardware-enforced confidential computing. The collaboration integrates AMD EPYC CPU performance, AMD Infinity Guard security features, and x86 software compatibility with Google Cloud’s TPU AI acceleration. This combination allows enterprises to deploy sensitive AI workloads—such as inference, fine-tuning, and reinforcement learning—in the cloud while maintaining robust data protection. The deployment reinforces AMD EPYC’s role as a foundational component for accelerated AI infrastructure across cloud environments.

In TPU 8tx systems, 5th Gen AMD EPYC processors handle workload orchestration, data preparation, and high-throughput communication with TPU accelerators. The system leverages AMD Secure Encrypted Virtualization (SEV) capabilities within the EPYC CPU, extending confidential computing protections to connected accelerators. This ensures customer prompts, model weights, and intermediate data remain encrypted in CPU memory and during transit between the host and accelerators, with hardware isolation during execution. The protections enable enterprises to confidently use sensitive data in reasoning and agentic AI applications.

AMD EPYC processors also support Google Cloud’s broader confidential computing portfolio, including Confidential VMs, Confidential GKE, Confidential Space, and Confidential Managed Service for Apache Spark. The TPU 8tx collaboration extends these protections to confidential AI acceleration, addressing a critical need for production AI: safeguarding data while it is in use. AMD and Google Cloud emphasize that the pairing of EPYC processors with TPU 8tx delivers silicon-level data protection alongside low-latency performance required for modern AI models.

AMD EPYC processors are the host CPU of choice for over 10 leading hyperscalers and frontier AI labs, supporting systems with AMD Instinct GPUs, Nvidia GPUs, and custom silicon. The breadth of compatibility across accelerator choices, combined with the established x86 software ecosystem, provides customers with a proven CPU platform for AI infrastructure. AMD and Google Cloud have collaborated across multiple generations of EPYC processors, including support for Google Cloud Compute Engine virtual machines, and have conducted deep reviews of AMD security processor firmware by Google security experts.