The upcoming NVIDIA Blackwell architecture, which will power the GeForce RTX 50 series GPUs, has been the subject of numerous rumors, and the latest leak pertains to the mobile SKUs. According to insider Moore’s Law is Dead, a photo surfaced on October 23rd showing a GPU that is expected to be featured in the RTX 5080 for laptops.
The insider claims that this GPU might have 8,196 CUDA cores and could be used in the standard RTX 5080 model or possibly a Ti or SUPER variant of the mobile graphics chip. The chip, labeled as “N22W-ES-A1,” is an engineering sample, indicating that it’s not the final version, but the designation aligns with earlier leaks, including one from Clevo, a notebook manufacturer.
Clevo’s August slide referenced the “GN22” GPU, rumored to be the core of NVIDIA’s high-end Blackwell-based mobile GPUs. These mobile chips are also expected to feature GDDR7 memory, marking an upgrade in-memory technology.
In terms of power consumption, the RTX 5090 mobile is rumored to retain the 175W TDP of the current RTX 4090 mobile model. The RTX 5080, however, is expected to be more power-efficient, reportedly dropping from 175W in its predecessor to around 140W, a significant reduction that could improve efficiency and thermal performance in laptops.
Entry and mid-range GeForce RTX 50
The recent Clevo leak also sheds light on the intermediate and entry-level SKUs within NVIDIA’s upcoming GeForce RTX 50 mobile series. For example, the RTX 5070 could see a reduction in power consumption by about 25W compared to the current RTX 4070 mobile variant. Meanwhile, entry-level models like the RTX 5060 are expected to maintain a 115W TDP, staying consistent with the previous generation.
One of the most significant advancements in the next-gen Blackwell architecture is the memory interface, particularly for the entry-level GPUs. These models will increase from a 96-bit to a 128-bit memory bus, providing a substantial boost in memory bandwidth. Combined with the shift to faster GDDR7 memory, this change is expected to offer a noticeable performance improvement over the GDDR6 memory currently used in Ada Lovelace-based GPUs.
The leaked image of the N22W-ES-A1 GPU suggests that NVIDIA is already sending engineering samples of these chips to partners, likely for the development of new-generation laptops. These notebooks could potentially hit the market as early as the first quarter of 2025, although no official confirmation has been made yet.
This timeline aligns with expectations for the desktop versions of GeForce RTX 50 cards, which are rumored to launch between January and March 2025, marking the arrival of NVIDIA’s latest GPU architecture in both mobile and desktop form factors.
