More Agents Per Watt, Per Rack
Scale your inference, agentic AI, and HPC with greater density at the same power. The AMD EPYC 9006 scales up to 256 cores; on legacy flagship 3rd gen processors, consolidate four to one (EPYC 7763 vs EPYC 9996).
Tightly Packaged Chiplet for Generational Uplift
AMD EPYC 9006 is still a chiplet, just more tightly packaged using TSMC’s N2, the first to do it. EPYC 9006's new architecture boasts 1.8x in performance and 1.5x in perf. per watt (9996 vs 9965).
Infrastructure that Fits Your Needs
Exxact Multi-GPU systems are balanced for your specific use case as your compute scales. Value your peace of mind with our systems backed by our 3-year warranty and support.
EPYC Powers Purpose-Built Servers for AI, HPC, and Enterprise
Built on cutting-edge Zen6 and Zen6c core architecture, AMD EPYC 9006 redefines what servers can achieve: more cores, more bandwidth, and more efficiency than ever before. EPYC 9006 features two new socket/platforms:
- SP7 EPYC 9006 & EPYC 9006X: Extreme processors engineered to push the envelope, prioritizing performance per core. Ideal for demanding cloud, HPC, and AI workloads
- SP8 EPYC 9006 & EPYC 9006LP: Versatile workhorse processors delivering top-tier performance while prioritizing cost efficiency and power draw. Ideal for storage, general purpose, and edge use cases.
Exxact TensorEX Servers featuring AMD's most ambitious platform yet, gives data centers and enterprises the balance of performance, density, and efficiency for every workload—from AI inference to large-scale virtualization to HPC simulation.
The Full Line-up of AMD EPYC 9006 “Venice” built on Zen 6
AMD EPYC 9006 SP7
Launching Q426. Designed for highest core density and leadership performance to support AI-first and cloud environments. Scale agentic workloads, virtual machines, and AI host nodes efficiently.
- Agentic AI Host Node
- AI Dev. Compute Node
- General Purpose Compute
AMD EPYC 9006 SP8
Launching 1H27. Deliver balanced compute, memory, and I/O to run business-critical applications, virtualization, and mixed enterprise workloads—alongside emerging agentic AI services.
- Agentic AI Host Node
- AI Dev. Compute Node
- General Purpose Compute
AMD EPYC 9006X SP7
Launching 2H27. Accelerate simulation, modeling, and memory‑intensive workloads with enhanced cache and bandwidth—enabling faster time-to-insight and supporting agent-driven analysis.
- HPC & Memory Intensive Compute
- AI Dev. Compute Node
AMD EPYC 9006LP (SP8)
Launching 2H27. Built for AI host node deployments, enabling efficient, scalable AI infrastructure with optimized performance per system and strong accelerator utilization.
- Efficient Low Power Head Node
- Edge AI and Compute Node
AMD EPYC 9006 coming soon Q426