NVIDIA Shifts RTX 5090 Chips to Professional GPUs

NVIDIA may be preparing to end production of its GeForce RTX 5090 as reports from hardware leakers and Taiwanese technology publication BenchLife indicate that the company has stopped supplying GB202 chips for the flagship gaming graphics card. The reported change would redirect production resources toward professional RTX PRO GPUs, potentially leaving a rumored 24GB RTX 5080 SUPER as NVIDIA's highest-end consumer graphics card still in production. According to BenchLife, NVIDIA has stopped supplying GB202 chips to graphics card partners for RTX 5090 production.

Hardware leaker MEGAsizeGPU similarly claims that future allocations of the GPU will be reserved exclusively for professional RTX PRO products. A separate report attributed to hardware leaker hongxing2020 describes an alleged factory notice stating that NVIDIA will no longer accept orders for RTX 5090 models, including the RTX 5090 D variants sold in China. The notice also indicates that the RTX 5080 16GB could eventually be discontinued in favor of a 24GB model.

However, the reported supply changes do not establish that NVIDIA has formally declared the RTX 5090 end-of-life. Existing chips and finished graphics cards could remain available through retailers, and NVIDIA has not confirmed a discontinuation date.

BenchLife reports that NVIDIA is targeting the first quarter of 2027 for a new RTX 5080 model with 24GB of memory. Board partners have reportedly received verbal notice of the planned product, although the publication says they have not yet received the development materials needed to prepare their designs. The rumored specifications align with earlier leaks about the GeForce RTX 5080 SUPER. The card is expected to use a GB203 GPU with 10,752 CUDA cores, 24GB of GDDR7 memory, a 256-bit memory interface, and a power limit of approximately 415W.

Those specifications would provide 50% more video memory than the standard RTX 5080 while retaining the same CUDA core count. NVIDIA has not confirmed the product's name, specifications, pricing, or release schedule. Previous reports attributed delays to rising GDDR7 memory costs.

The potential shift toward professional GPUs comes as NVIDIA serves demanding artificial intelligence, workstation, and visualization workloads. Its RTX PRO 6000 Blackwell uses the GB202 GPU and features 24,064 CUDA cores alongside 96GB of GDDR7 memory.

The RTX PRO 5500, another professional model, features 21,760 CUDA cores, matching the RTX 5090's core count, and includes 84GB of memory. Professional graphics cards generally command substantially higher prices than consumer GeForce products, although NVIDIA has not publicly linked any RTX 5090 supply changes to a strategy of prioritizing professional GPUs. The rumored RTX 5080 SUPER would also leave a substantial performance gap if it becomes the highest-end GeForce card in production.

In The PC Enthusiast's 13-game benchmark suite, the RTX 5090 averaged 169 frames per second at 4K resolution, compared with 115 FPS for the RTX 5080, a difference of approximately 47%. The leaked RTX 5080 SUPER specifications suggest that its main advantages over the standard RTX 5080 would be increased memory capacity and potentially higher clock speeds rather than a larger CUDA core count. An additional 8GB of GDDR7 could benefit memory-intensive games, content creation, and AI workloads, but capacity alone does not guarantee substantially higher gaming frame rates.

RTX 5090 availability could tighten further if NVIDIA has stopped allocating new GB202 chips to consumer graphics cards. The PC Enthusiast reported in September that limited availability had pushed some third-party marketplace asking prices close to $10,000.

Remaining inventory could become more expensive if supplies decline, although existing stock may continue reaching retailers. NVIDIA has not confirmed the reported production changes, announced an official end-of-life date, or provided a confirmed launch date for the rumored 24GB RTX 5080 replacement.