Later this year, AMD will release its next-generation laptop processors, which will coincide with the launch of the new Ryzen 9000 desktop processors. These new processors represent a significant shift in AMD’s approach to laptop CPU model numbers.
A key focus for AMD is the enhanced performance in generative AI and machine learning workloads, prominently featuring an upgraded neural processing unit (NPU) that surpasses Microsoft’s performance standards for Copilot+ PCs. The new Ryzen AI 300-series, known as Strix Point, follows the Ryzen 8040 chips from earlier this year, which themselves were minor updates to the Ryzen 7040 processors from less than a year prior.
AMD claims its new third-generation NPU can deliver up to 50 trillion operations per second (TOPS), a substantial increase from the 10 to 16 TOPS of the Ryzen 7000 and 8000 processors with NPUs. This performance surpasses the 45 TOPS of the Qualcomm Snapdragon X Elite and X Plus in the initial Copilot+ compatible PCs, as well as Intel’s expected performance for its upcoming Core Ultra chips, codenamed Lunar Lake. All these processors exceed Microsoft’s Copilot+ requirement of 40 TOPS, which enables certain Windows 11 features typically unavailable on standard PCs. Copilot+ PCs can perform more AI processing locally, enhancing performance and privacy by reducing reliance on cloud services.

For users uninterested in generative AI, the Ryzen AI 300 processors still offer significant updates, including a CPU based on the Zen 5 architecture and an “RDNA 3.5” integrated GPU to enhance gaming performance for slim laptops that lack a dedicated graphics processor. These chips are produced using TSMC’s N4 process.
AMD is introducing two high-end Ryzen 9 series chips. The Ryzen AI 9 HX 370 features 12 CPU cores and 16 GPU cores, compared to the Ryzen 8040 series’ maximum of eight CPU cores and 12 GPU cores. The Ryzen AI 9 365 offers 10 CPU cores and 12 GPU cores. Both chips include the same NPU.
Although an increase in CPU core count suggests significant improvements in multi-threaded performance, it’s worth noting that many of these cores (8 in the 370, 6 in the 365) use the “Zen 5c” architecture, a compact variant of Zen 5 optimized for small size rather than high clock speeds. This is similar to Intel’s E-cores but without the heterogeneous CPU architecture that has caused compatibility issues with some apps and games.
This is the first time AMD has used a mix of big and small CPU cores at the high-end, with Zen 4c cores previously appearing only in lower-end, lower-power CPU designs in the Ryzen 3, 5, and Ryzen Z1 families.
Interestingly, AMD did not provide direct performance comparisons between the Ryzen AI 300 chips and the Ryzen 8040 series, instead focusing on comparisons with Intel, Qualcomm, and Apple offerings. This could suggest that the generational improvements are not as substantial as AMD would like to emphasize.
Similarly, AMD did not compare the new Radeon 890M and 880M to the older Radeon 780M. They did state that the 890M is on average 36 percent faster in select games compared to Intel’s Arc integrated GPU in the Meteor Lake Core Ultra chips and 60 percent faster than the Snapdragon X Elite in the 3DMark Night Raid benchmark. However, most games on Snapdragon PCs still face the overhead of code translation from x86.
AMD anticipates that Ryzen AI chips will be available in over 100 platforms from OEMs starting in July 2024, shortly after the release of Microsoft and Qualcomm’s first wave of Snapdragon X-equipped Copilot+ PCs. Ryzen AI will also compete with Intel’s upcoming Lunar Lake chips, expected later this year.