AMD Ryzen and Radeon Next-Gen Hardware Explained

AMD has revealed details about its next-generation desktop platform through new entries in its technical information portal. The company now lists both Olympic Ridge and Medusa 1. Olympic Ridge has been leaked as AMD's next-generation desktop Ryzen platform, and the listing shows AM5 support. That means the next generation will be supported on the current platform.

Modern PC security, future GPU architecture, and graphics card pricing are all changing at once. AMD has confirmed vulnerabilities in TPM 2.0 reference code, and Intel systems are affected as well. At the same time, new information points to changes in how future Radeon GPUs could be designed, while graphics card prices are beginning to rise again.

Zen 6 Could Bring More Cores and Higher Clocks

Medusa 1, also known as Medusa Point, is set to be AMD's Zen 6-based mobile processor platform. It will be supported on both FP10, the mobile socket, and AM5. This points toward next-generation desktop APUs coming to the platform. AMD has not announced motherboard compatibility, so it is unclear whether every existing model will be supported.

2027 is shaping up to be a year focused on AMD's next-generation desktop CPUs and gaming GPUs. Ryzen 10000, or whatever the processors end up being called, is based on AMD's next-generation Zen 6 architecture, which is confirmed to be built on TSMC's 2nm node. That represents two node jumps from Zen 5's 4nm process.

Leaks are pointing to a 10%+ IPC increase over Zen 5. That would boost performance, but higher clocks could add another part to the gain. Moore's Law Is Dead has claimed that AMD could push Zen 6 processors to as high as 7 GHz. A source has also claimed to have seen documentation indicating that eventual qualification samples reached that clock speed.

Future Radeon GPUs Could Change Cache Design

While AMD fixes security problems on current hardware, clues also suggest what the company may be planning next. Future Radeon GPUs could look different from current designs. For years, one of AMD's GPU technologies has been Infinity Cache. The basic idea is simple. Instead of constantly accessing the graphics memory, the GPU can keep frequently used data closer to the chip.

That can improve efficiency and help reduce memory bandwidth pressure. Now, AMD appears to be changing that strategy. Newer AMD architectures are moving toward much larger shared L2 caches, which could be an important change. A larger, more efficient L2 cache can reduce latency, cut unnecessary memory accesses, and improve power efficiency.

AMD is already using this kind of design in newer compute-focused GPUs. Leaks suggest that similar technology could eventually appear in future gaming GPUs as well, which fits with AMD's broader plan. The company has been moving toward bringing its gaming and compute GPU technologies closer together.

Technology developed for AI and data center GPUs could eventually benefit Radeon gaming cards, too. Instead of maintaining completely separate architectures, AMD could build a more unified design and then scale it across products, including gaming, AI, workstations, and servers, with more of the same underlying technology.

The highest-end chip could reportedly receive a 50% increase, reaching up to 24 cores. More cores are also expected throughout the product stack, as the number of chiplets that make up the CPUs is reportedly increasing. The changes could also affect AMD's X3D processors. The next generation could allow up to 12 cores on a single chiplet.

Cross-CCX communication can introduce higher latency, so more cores communicating on the same chiplet could help. Zen 6 X3D is also expected to feature up to 96 MB of L3 V-Cache instead of 64MB. Multiple leaks also claim that AMD is working on new technologies to improve 1% lows in games on next-generation Ryzen processors.

Moving to GPUs, AMD's RX 10000 series is expected to be based on RDNA 5 or potentially UDNA, representing AMD's upcoming unification of its RDNA gaming architecture and CDNA compute architecture. One leak claims that RDNA 5 is based on UDNA with additional changes for the PS6.

The architecture may still be called RDNA 5, but according to the leak, it would still be based on UDNA.

The leaked chip is identified as AT0, which is the actual die that makes up the GPU. Another well-known leaker, Kepler, has claimed that if AMD releases the AT0 die, it will be released to consumers. The same leaked information points to a GPU with 36GB of GDDR7. A Radeon GPU with that configuration could potentially challenge NVIDIA's 6090 at the high end.

AMD has also confirmed technologies from Project Amethyst that will appear in future hardware. One of them is a new dedicated hardware block called the Radiance Core, designed specifically for ray traversal. That could significantly improve ray-tracing performance in future Radeon GPUs. AMD is also working on neural arrays for neural rendering and universal compression.

These technologies suggest AMD is reworking how its next-generation GPUs handle ray tracing, AI, and memory bandwidth. If the current leaks and AMD's confirmed technologies align, 2027 could be a major year for AMD's CPU and GPU hardware.