AMD is rumoured to be preparing to bring its Zen 7 architecture, codename “Grimlock”, into development soon. On top of that, it is expected to be the first architecture to be fabricated around the 14A process, which will also make it the Zen architecture to enter the Angstrom node era.
Reports of AMD prepping beyond Zen 6 were first reported by Taiwanese news portal, Commercial Times. The site says that Lisa Su, CEO of AMD, has been on frequent visits to the island country, presumably to secure chips from TSMC and more specifically, chips from its 14A node. This is on top of its already purchased stock of 2nm for its Zen 6 architecture.
Still on Zen 7, the same report also suggests that AMD could be considering using Powertech Technology’s FOPLP packaging. That’s short for Fan-out panel-level packaging, and it’s used to prepare larger and more complex chiplet packages.
The report also makes mention of the next generation 3D V-Cache for Zen 7, which isn’t all that surprising. The stacked cache technology has come a long way since its debut with the Ryzen 5000 Series. In the case of Grimlock, the CCD is rumoured to feature up to 16 cores, while the 3D V-Cache variants could reach up to 224MB per CCD.
But that’s what those details are at this point: rumours. Zen 7 is still a full CPU generation away, and AMD still has Zen 6 to launch. To that end, other reports and rumours suggest that the CPU and GPU maker will provide more details about the CPU architecture during its keynote at Computex 2026, which is happening next week. That architecture is expected to scale up to 32-cores, and have 3D V-Cache variants with L3 Caches of up to 288MB on two chiplets. Effectively bringing the fight to Intel and its own upcoming Nova Lake architecture, which itself is expected to have CPUs that scale up to 52-cores for its top-spec SKU.
(Source: Commercial Times, TPU)
