Advanced Micro Devices debuted the Epyc 6 server processor series, code-named Venice, delivering up to 256 cores on the new Zen 6 microarchitecture.
The lineup arrives as artificial intelligence agents shift computational demands across enterprise datacenters. According to Nextplatform, the emerging agentic AI market is forecast to expand rapidly, with some analyst firms calling for it to grow almost 43 percent annually between now and 2034.
Unlike conversational chatbots that rely almost exclusively on graphics processing units, AI agents break tasks into multi-step workflows, execute tools, query databases, and manage application programming interfaces. Dan McNamara, senior vice president and general manager for compute and enterprise AI at AMD, described the shift in an article published in May.
the shape of the workload changes entirely,
Dan McNamara, senior vice president and general manager for compute and enterprise AI at AMD
AI is significantly expanding the role of the CPU across every layer of the data center,
Dan McNamara, senior vice president and general manager of AMD’s Compute and Enterprise AI unit
The Epyc 6 processors comprise 10 distinct central processing units split into high-core-count and high-frequency categories, as reported by Siliconangle. The top-tier densest chips feature 256 cores with a base speed of 2.55 gigahertz, capable of supporting 512 threads to manage concurrent agent workloads. Meanwhile, the highest-frequency models run a base speed of 4.05 gigahertz and boost up to 5.3 gigahertz with 96 cores and a 384-megabyte cache.
Underpinning the silicon is Taiwan Semiconductor Manufacturing Co.’s two-nanometer foundry node, enabling a Zen 6 instruction-per-clock performance increase of 31 percent in AI workloads over previous designs, according to Siliconangle. TechPowerUp noted that the performance core complex die contains 12 Zen 6 cores sharing 48 megabytes of L3 cache, representing a 50 percent increase in core and cache counts.
Mike Clark, corporate fellow, senior vice president, and lead Zen architect at AMD, reflected on the engineering trajectory during a briefing with analysts.
we didn’t have a foresight that agentic was going to be what it has become, but we have always believed that the Core Complex Die was going to play a critical role in AI. We did a big investment in Zen 5 to do that, but with Zen 6, we continued to invest down that path and enhance our AI capability and it’s paying off as agentic has taken off.
Mike Clark, Corporate Fellow, senior vice president, and lead Zen architect at AMD
Madhu Rangarajan, corporate vice president for compute and enterprise AI products, detailed how the company evolved its product family following the introduction of Naples chips.
So with your Genoa and your Turin and your Bergamo and your Siena, you had your optimized design points going after different sub-segments of the market,
Madhu Rangarajan, corporate vice president for compute and enterprise AI products
Rangarajan also emphasized the breadth of the current hardware ecosystem.
That’s where we really started growing up to have a full portfolio of products here, and continued that.
Madhu Rangarajan, corporate vice president for compute and enterprise AI products
The platform supports high-speed memory bandwidth across channels of DDR5 memory and introduces support for PCIe Gen 6, according to TechPowerUp. Benchmarks provided by AMD indicate that a 256-core configuration delivers up to 3.3 times more agents-per-watt than Nvidia Corp.’s Vera CPU, as noted by Nextplatform.
What agentic AI is doing is it’s driving CPU demand across three networks vs. just general-purpose servers,
Madhu Rangarajan, corporate vice president for compute and enterprise AI products
Agents like to do searches, in search clusters. Agents execute enterprise tools – CRM, ERP – we increasingly have agents doing EVM workloads. Your agents are accessing databases. All of this needs a broad CPU portfolio for these workloads.
Madhu Rangarajan, corporate vice president for compute and enterprise AI products
Ephemeral tools are the more interesting part of agentic AI because there is no well-defined benchmark there,
Madhu Rangarajan, corporate vice president for compute and enterprise AI products
Hardware partners are shipping SP7 server systems globally. Microsoft Azure announced a preview of its HDv2 instances built on Venice, making it the first cloud provider to offer the 6th Generation Epyc architecture, alongside upcoming deployments on Oracle Cloud Infrastructure, as reported by Technetbooks and TechPowerUp.
Earlier coverage: AMD Records $1.07 Billion Charge for AI Deals with OpenAI, Meta