Anticipation is building ahead of CEO Jensen Huang’s keynote speech at CES on January 6, where he is expected to unveil the next-generation Blackwell GPU architecture. NVIDIA is even resurrecting its GeForce LAN event after 13 years of absence, scheduled for January 4, suggesting important announcements.
According to a new report from Hong Kong technology media HKEPC, NVIDIA’s highly anticipated RTX 5080 graphics card should arrive on shelves on January 21, two weeks after its planned presentation at CES 2025. This schedule matches NVIDIA’s release habits, which generally leaves a short gap between announcement and commercial availability.
The technical specifications of the RTX 5080 promise, unsurprisingly, to be very high-end. The card would feature the GB203-400-A1 chip with 10,752 CUDA cores distributed across 84 streaming multiprocessors, an increase of 10% compared to its predecessor. While retaining the 16GB memory configuration of the RTX 4080, the new card upgrades to GDDR7 memory technology clocked at 30Gbps, notably faster than other planned RTX 50 cards, which are expected to use 28Gbps modules.
Also read – The Nvidia RTX 5090 and 5080 will introduce a new function, this leak has just revealed everything
The RTX 5080 will be a real power monster
Manufacturing improvements will play a key role in the new architecture. Blackwell GPUs will use the TSMC custom 4NP process nodeproviding 30% greater density than the 4N node used in current Ada Lovelace cards. Even if Nvidia didn’t go with the most advanced 3nm, it should yield significant performance gains beyond just increasing the number of CUDA cores.
-Interestingly, the launch of the RTX 5080 seems to precede that of the RTX 5090breaking with NVIDIA’s traditional release strategy. The timeline for other cards in the lineup, including the RTX 5070 series, remains unclear.
With AMD currently absent from the high-end GPU segment, NVIDIA faces little competitive pressure in pricing the RTX 5080. However, final pricing details will only be confirmed during the CES announcement.
Although HKEPC has a fairly reliable history, it should be noted that this information remains unofficial until NVIDIA’s formal announcement. The January 21 date is specific to the RTX 5080, as the launch schedule for other Blackwell GPUs may vary. As a reminder, this split launch strategy has already been used by Nvidia with previous generations.
