AMD Zen 6 “Olympic Ridge” leaks: The 8-core era is finally over

For years, AMD kept things consistent, almost stubbornly so. Since Zen 2, every desktop CCD (Core Complex Die) packed exactly 8 cores.

That was the rule. Zen 3 came, Zen 4 came, Zen 5 came, 8 cores per chiplet, no questions asked. Well, that structure is finally changing with Zen 6, which is set to move from 8 to 12 cores per CCD, and from 32MB to 48MB of shared L3 cache, a 50% jump in both figures.

The leak comes from well-known hardware insider HXL, and by now it’s been confirmed by multiple sources. The codename for AMD’s next-gen desktop lineup? Olympic Ridge.

12 cores, same size, how is that even possible?

That’s the part that genuinely impresses. You’d expect that cramming 50% more cores into a chiplet would make the die noticeably larger, but that’s not what’s happening here.

AMD’s engineers managed to pack 12 full-fat cores into a single CCD thanks to a shift to TSMC’s 2nm process node, yet the area of the new 12-core die has remained almost the same as its predecessors, around 76 square millimeters, compared to 71 for Zen 5. More cores, barely any extra silicon. That’s what a node jump can do.

This is a major change over AMD’s norms for Ryzen CPUs, which have used 8-core CCDs since Zen 2. It’s worth noting though, that AMD’s Zen 2 CCDs were actually configured as two 4-core clusters with 16MB of L3 cache each, it wasn’t until Zen 3 that AMD unified those into a single shared L3 pool. Zen 6 takes that foundation and blows it wide open.

What this means for the Olympic Ridge lineup

According to the post from HXL, single-chiplet options would include 6, 8, 10, and 12 cores. The dual-chiplet parts listed are 16 cores (8+8), 20 cores (10+10), and 24 cores (12+12). Each CCD will offer up to 12 Zen 6 cores paired with 48MB of L3 cache per chiplet, which would make a dual-CCD desktop part a 24-core, 96MB L3 configuration, before any 3D V-Cache variants.

And speaking of X3D, a future Ryzen 7 X3D chip could boast 12 cores and a massive 144MB of total L3 cache, combining the 48MB native L3 with 96MB of stacked V-Cache. That’s a number that would make any gamer’s eyes light up.

On top of all this, AMD is reportedly targeting 10%+ IPC gains and clock speeds above 6GHz, thanks to TSMC’s N2X node, a refined version of N2 built for higher frequencies.

AMD hasn’t officially confirmed any of these specs yet, and neither AMD nor Intel has confirmed this data. But if the information is accurate, we’re in for a heated second half of 2026, where both platforms will get processors that will define the look of PCs for years to come.

The 8-core era had a great run. Looks like it’s time to move on.

What do you think about AMD finally breaking the 8-core-per-CCD mold? Are you excited for Olympic Ridge or still holding out to see real benchmarks? Drop your thoughts below!