AMD Unveils EPYC Verano and Venice-X Server CPUs for 2027
AMD has introduced two next-generation server processors, EPYC Verano and EPYC Venice-X, during its Advancing AI 2026 keynote. Both chips are scheduled for production in 2027 and extend the Zen 6 architecture into specialized workloads, including agentic AI and high-performance computing (HPC). Moreover, the new processors target different performance needs instead of relying on a single server design.
Venice-X Delivers Massive Cache for High-Performance Computing
EPYC Venice-X features 96 Zen 6 cores alongside AMD’s 3D V-Cache technology. As a result, it offers 1,152 MB of L3 cache beneath the compute dies while reaching clock speeds of up to 5.15 GHz.
Additionally, the processor supports 16-channel memory with DDR5-8000 RDIMMs or MRDIMMs operating at 12,800 MT/s. Consequently, it delivers up to 1.6 TB/s of aggregate memory bandwidth, nearly three times higher than the current Turin-generation EPYC processors.
Microsoft has confirmed it will deploy Venice-X in future Azure HXv2 virtual machines for electronic design automation and other bandwidth-intensive workloads. Meanwhile, Venice-X uses the SP7 socket shared with the broader Venice family, which is already in production on TSMC’s 2nm process. While standard Venice processors scale up to 256 cores for cloud computing, Venice-X prioritizes cache capacity and higher frequencies for technical workloads that benefit from faster data access.
Verano Focuses on Agentic AI Infrastructure
EPYC Verano, part of the EPYC 9007 family, takes a different approach by offering up to 72 cores with SOCAMM2 memory based on LPDDR5X modules. Therefore, the processor improves power efficiency for dense AI inference racks while maintaining strong performance.
Furthermore, Verano supports up to 112 Gbps xGMI connectivity to pair with AMD’s upcoming Instinct MI500 GPUs in next-generation agentic AI systems. The processor also runs on the new SP8 platform instead of the SP7 infrastructure used by Venice.
AMD designed Verano specifically for autonomous AI agents that require sustained, low-latency access to large working data sets.
AMD Eyes Expanding AI Server CPU Market
The latest processor launches align with AMD’s updated server CPU market outlook. The company now expects the total addressable market to exceed $120 billion by 2030, compared with its earlier estimate of about $60 billion. In addition, CEO Lisa Su said during AMD’s May earnings call that the market is expanding at a compound annual growth rate exceeding 35 percent.
Therefore, AMD continues expanding its EPYC portfolio with frequency-optimized, cache-focused, and energy-efficient processors to address the rapidly growing demand for AI inference and specialized computing workloads.
