Nvidia will release RTX 5090 and RTX 5080 GPUs later this year

According to data from Wccftech, Nvidia plans to introduce new GPU models - RTX 5090 and RTX 5080 - at the end of 2024. These products will be aimed at the high-performance computing and artificial intelligence segment.

In addition, the next generation of RTX graphics will be powered by Blackwell GPU IP, using the RTX 50 series branding.

The new graphics cards are expected to help Nvidia's partners increase shipments and raise the average selling price of GPUs as more gamers opt for high-quality service. With the new GeForce RTX 50 series and related AI-powered PC platforms, major brands are expecting significant growth.

Manufacturers expect that Nvidia intends to release two flagship chips called GB202 and GB203. The first one will be used in the RTX 5090, and the second one in the RTX 5080.

Experts suggest that the memory configurations on various Blackwell GPUs will look like this:

• GB202 – 512-bit / 28 Gbps / 32 GB (maximum memory) / 1792 GB/s (maximum bandwidth);
• GB202 – 384-bit / 28 Gbps / 24 GB (maximum memory) / 1344 GB/s (maximum bandwidth);
• GB203 – 256-bit / 28 Gbps / 16 GB (maximum memory) / 896.0 GB/s (maximum bandwidth);
• GB205 – 192-bit / 28 Gbps / 12 GB (maximum memory) / 672.0 GB/s (maximum bandwidth);
• GB206 – 128-bit / 28 Gbps / 8 GB (maximum memory) / 448.0 GB/s (maximum bandwidth);
• GB207 – 128-bit / 28 Gbps / 8 GB (maximum memory) / 448.0 GB/s (maximum bandwidth).

Nvidia Blackwell gaming GPUs will use the TSMC 4NP node. This is expected to significantly improve SM 1's cache and throughput.