AMD EPYC 9965 192-Core Processor (Turin Dense)

The AMD EPYC 9965 is the flagship density-optimized processor in AMD's fifth-generation EPYC 'Turin Dense' family, featuring an unprecedented 192 cores and 384 threads built on the Zen 5c microarchitecture. Fabricated using TSMC's 3nm process node, the processor integrates multiple compute chiplets alongside a central I/O die, delivering extreme thread counts within a 500W TDP envelope suited to high-density rack deployments. With 384MB of L3 cache and support for 12-channel DDR5 memory per socket, the EPYC 9965 sustains high memory bandwidth even under fully saturated multi-tenant workloads.

The Zen 5c core is a physically compact variant of the Zen 5 architecture, retaining the full Zen 5 instruction-set capabilities—including AVX-512 with 512-bit FPU execution—while enabling a higher core count per die. The processor supports up to 6TB of DDR5 RAM per socket using registered DIMMs and exposes 160 PCIe 5.0 lanes, making it capable of driving dense NVMe storage arrays and high-bandwidth networking fabrics simultaneously. Security features include AMD Infinity Guard technologies such as Secure Encrypted Virtualization-Secure Nested Paging (SEV-SNP), supporting confidential computing deployments at hyperscale.

Positioned as the highest-core-count x86 server processor available at its launch in Q4 2024, the EPYC 9965 targets hyperscale cloud operators, HPC centers, and large enterprise virtualization hosts that prioritize VM or container density per rack unit above all other metrics. Its SP5 socket compatibility with the broader EPYC 9004 platform ecosystem allows operators already invested in Turin or Genoa infrastructure to deploy the 9965 with qualified platform support from major server vendors.

Ideal for

• Hyperscale cloud infrastructure requiring maximum virtual machine density per 1U or 2U server node
• High-performance computing clusters running tightly parallel workloads such as molecular dynamics, climate modeling, or financial Monte Carlo simulations
• Large-scale containerized Kubernetes environments where CPU thread count per host directly reduces node sprawl
• Enterprise virtualization consolidation on VMware vSphere or Red Hat OpenShift, maximizing vCPU-to-core ratios across fewer physical servers
• In-memory analytics platforms such as SAP HANA scale-up instances leveraging the processor's high DDR5 memory capacity support
• Confidential computing and multi-tenant cloud hosting utilizing AMD SEV-SNP to isolate workloads at the hardware level

Technical specifications

Available from Omnixon Global. Submit an RFQ and our team will confirm configuration and availability for your order.