NVIDIA appears to be lining up a second version of its GeForce RTX 5050, and the big change is an unusual memory configuration: 9GB of VRAM. The claim comes from leaker MEGAsizeGPU, who has regularly posted accurate NVIDIA-related SKU details in the past, making this one worth watching even though nothing is official yet.
If the information holds, the new card will still be called GeForce RTX 5050, but it won’t match the existing 8GB model. Instead, it’s said to ship with 9GB of GDDR7 memory on a narrower 96-bit memory bus, replacing the 128-bit bus found on the current RTX 5050 8GB. That mix of a smaller bus and slightly higher VRAM is a notable shift for an entry-level graphics card, especially because most budget GPUs tend to stick to more conventional 8GB or 12GB configurations.
The technical trade-off is where things get interesting. The reported 9GB RTX 5050 would use faster 28Gbps GDDR7 memory, which could push total memory bandwidth to around 336GB/s. By comparison, the existing 8GB model uses 20Gbps memory and lands at roughly 320GB/s. On paper, that’s about a 5% bandwidth bump, alongside a 12.5% increase in VRAM capacity. In real gaming workloads, that extra gigabyte may help in a handful of newer titles that are sensitive to VRAM limits at 1080p with high textures, though the narrower 96-bit bus could still affect performance in bandwidth-heavy scenarios.
Core specifications are expected to stay the same. The rumored RTX 5050 9GB would reportedly continue to use the Blackwell-based GB207 GPU, described as the smallest chip in the Blackwell gaming stack. The leak points to 2,560 CUDA cores, matching the core count of the older RTX 3050 but paired with the newer Blackwell architecture. Power is also expected to remain unchanged, with a 130W board power and a standard 8-pin power connector.
Pricing and launch timing haven’t been confirmed. The existing RTX 5050 8GB has been associated with a $249 price point, and there’s a reasonable chance NVIDIA could keep the 9GB model in the same range to avoid complicating the entry-level lineup—although final pricing will likely depend on supply, partner costs, and competitive pressure.
The same leaker also claims NVIDIA may adjust its GeForce RTX 5060 plans. According to the report, RTX 5060 cards could be produced using GB205 GPU dies rather than the expected GB206, allegedly due to limited availability of GB206 supply. The idea is that board partners could be given cut-down or defective GB205 chips to meet volume needs for mainstream models. If that happens, it could force add-in-board partners to create new PCB designs to accommodate the different GPU layout, effectively adding cost and complexity—especially if the design ends up resembling higher-tier boards while still using an 8-pin connector.
All of this is unfolding against a backdrop of continued GPU availability challenges. The report suggests board partners are juggling pressure across multiple models, particularly in the entry-level and midrange segments like the RTX 5050 and RTX 5060, while other cards are also impacted by supply constraints. There’s also an indication that older-generation options such as the RTX 3060 may continue to be produced or emphasized to shore up demand where newer inventory is tight.
For buyers, the potential RTX 5050 9GB is worth keeping on the radar if you’re shopping for a budget NVIDIA GPU aimed at 1080p gaming. More VRAM and slightly higher bandwidth could be meaningful for certain games and settings, but final performance will depend on clocks, real-world memory behavior on a 96-bit bus, and—most importantly—how NVIDIA and its partners price and position the card once it’s officially announced.
