The GeForce RTX 5080 Super from NVIDIA retains 10,752 CUDA Cores while gaining 24 GB of GDDR7.
The dependable leaker kopite7kimi has revealed new rumors regarding the impending GeForce RTX 5080 SUPER, which suggests that NVIDIA’s next “SUPER” update may be on the horizon when Computex 2025 gets underway. The leaker revealed in a recent post on X that the new card will still use the GB203 GPU, matching the 10,752 CUDA cores of the current RTX 5080. The speed and memory capacity are the only SUPER improvements. The SUPER variation has 24 GB of GDDR7 @ 32 Gbps, which gives it a theoretical bandwidth of 1 TB/s, in contrast to the regular model’s 16 GB of GDDR7 operating at 30 Gbps. High-resolution gaming and professional workloads that require large frame buffers may benefit greatly from this increase of VRAM. Early estimates of bandwidth indicate a 6% increase in throughput over the 960 GB/s of the regular RTX 5080.
Denser 3 GB modules, such those found in the RTX PRO 6000, are necessary to achieve the increased memory density. As a result, power needs will increase. According to rumors, the SUPER model draws more than 400 W, which is roughly 40 W more than the 360 W rating of the vanilla RTX 5080. Now, aside from the previous generation top-tier RTX 4090, only NVIDIA’s flagship RTX 5090—which has 21,760 cores and 32 GB of GDDR7 across a 512-bit bus—will surpass the 5080 SUPER in terms of compute and memory resources. The precise launch date has not yet been established. Although previous speculation suggested a Q4 2025 release, a revised timeline might see NVIDIA introduce the new card at a significant trade show like CES in early 2026. Equally speculative is the pricing. It would be positioned to avoid direct rivalry with NVIDIA’s entry-level professional Blackwell GPUs, with an MSRP target of $1,000 to $1,500. As usual, specs and launch dates should be regarded with caution until formal announcements are made.
