A new leak tied to Weibo suggests Qualcomm’s next flagship chip may prioritize graphics and AI over raw CPU speed. The chipset, believed to be the Snapdragon 8 Elite Gen 6 Pro (SM8975), is said to deliver less than a 20 percent real-world CPU performance uplift.
The information comes from a user known as “Fixed-focus digital cameras”. While unverified, the claims align with a broader trend across the industry: CPU gains are flattening, while GPU and AI accelerators take center stage.
CPU gains appear limited despite the 2nm transition
According to the leak, the SM8975 will use a 2+3+3 core configuration built on TSMC’s GAA-based 2nm process. On paper, that node shift should bring efficiency and density improvements. In practice, the reported CPU uplift remains under 20 percent.
That figure likely reflects real-world mixed workloads rather than synthetic peak scores. It also suggests Qualcomm may be dialing back aggressive CPU scaling in favor of thermal stability and sustained performance.
This wouldn’t be unusual. Recent flagship chips have increasingly hit diminishing returns in CPU-bound tasks, especially in smartphones where power limits are strict.
GPU and memory changes take priority
The more interesting claim centers on the new A850 GPU. The leaker points to significant cache increases, which typically improve bandwidth efficiency and reduce latency in graphics-heavy scenarios.
Support for LPDDR6 memory is also mentioned. That would mark a generational jump in memory throughput, feeding both GPU workloads and on-device AI models with higher data bandwidth.
The emphasis here is clear: graphics rendering, AI inference, and high-throughput tasks are becoming the primary battleground.
Qualcomm’s direction shift
Qualcomm has already shifted its messaging around “Elite” branding toward sustained performance and AI capabilities rather than peak benchmark wins.
The Snapdragon 8 Elite Gen 3 and Gen 4 generations leaned heavily into NPU gains and efficiency improvements. If this leak holds, Gen 6 Pro continues that trajectory, with GPU and memory subsystems doing more of the heavy lifting.
Apple and MediaTek are making similar moves. CPU improvements are becoming incremental, while GPU and AI blocks see larger architectural changes.
What this means for real-world use
For users, a sub-20 percent CPU boost may not translate into noticeable day-to-day differences. App launches, UI navigation, and general responsiveness are already near-instant on current flagship chips.
Gaming, however, could benefit more directly. A larger cache and faster memory subsystem can improve frame stability and reduce stutter, especially in demanding titles.
AI workloads may see the biggest shift. Higher memory bandwidth and cache improvements are critical for running larger on-device models, including generative AI tasks.
That raises a different question. If CPU scaling slows while GPU and AI accelerate, future flagship comparisons may hinge less on Geekbench scores and more on sustained gaming performance and AI throughput—metrics that are still harder to standardize across devices.
