Home / Component / CPU / AMD Zen 6 CPUs will reportedly be built using TSMC N2P and N3P

AMD Zen 6 CPUs will reportedly be built using TSMC N2P and N3P

The blueprint for AMD's next generation of CPUs is beginning to take shape, with a new leak providing a detailed look at the process nodes that will work as the foundation for the upcoming Zen 6 architecture.

According to Kepler_L2 (via 3DCenter), AMD has shared its process node plans with partners, revealing a strategy that leans heavily on TSMC's advanced N2P node for high-performance parts, while utilising N3P for more mainstream and mobile products. Starting with the server segment, both the standard EPYC “Venice Classic” and the high-density “Venice Dense” processors are slated to use TSMC's N2P node. This aligns with previous rumours that AMD is looking to increase core counts, with the 2nm-class node providing the necessary transistor budget for the rumoured 12-core “Classic” and even 32-core “Dense” Zen 6 chiplets.

Desktop gamers and high-end chips will also probably be based on the same node as their server counterparts. The next-generation Ryzen platform, codenamed “Olympic Ridge”, is also set to have its core chiplets built on the N2P node. This will extend to high-end laptops, with the “Gator Range” platform also leveraging N2P for higher-tier mobile solutions.

Things get more mixed in the mainstream laptop segment with the “Medusa Point”, the successors to the current “Strix Point” family. Here, AMD will reportedly employ a mixed approach. For the top-tier Medusa Point SKUs, the processor will be a chiplet design, combining an N2P-based CCD with an I/O die (IOD) manufactured on the N3P node. Further down the stack, the lower-end Medusa Point SKUs are expected to be monolithic chips built entirely on the N3P process.

While this latest leak provides a great deal of clarity on the CPU side, other codenames like the high-end mobile “Medusa Halo” and low-end “Bumblebee” are also being mentioned, though details are scarce. As for the next-generation RDNA5 graphics architecture, sometimes referred to as UDNA, its process node remains unconfirmed.

Discuss on our Facebook page, HERE.

KitGuru says: What are your expectations for the next generation of Zen-based processors?

Become a Patron!

Check Also

Intel Arrow Lake Refresh has been benchmarked in Geekbench

New data from Geekbench has surfaced for Intel's unannounced "Arrow Lake Refresh", branding the upcoming performance bump as the "Plus" series. While these chips weren't mentioned during Intel's CES 2026 keynote, the new leaks suggest that both desktop (LGA-1851) and mobile (HX) refreshes are right around the corner...