Earlier than anticipated, Zotac's official website showed off the new GeForce RTX 5090, RTX 5080, RTX 5070 and RTX 5070 Ti models, confirming the memory specs selected by Nvidia.
Starting with the flagship RTX 5090 flagship, it's now certain that the successor RTX 4090 will be equipped with 32GB of GDDR7 memory, marking a significant upgrade that hints at a considerable jump in performance. According to recent rumors, the RTX 5090 could be up to 70% faster than the RTX 4090, making it an attractive choice for high-end hardware enthusiasts willing to forgo the previous generation for a major upgrade.
On the other hand, the RTX 5090 is rumored to be priced significantly higher than its predecessor, justified by the improved performance. In addition, speculators are expected to try to monopolize the initial stocks, hoping to capitalize on the high demand from enthusiastic users and those involved in the AI industry, where the demand for top graphics cards is constantly growing.
More accessible to the gaming community, the GeForce RTX 5080 comes equipped with the same amount of memory as the RTX 4090, 24 GB GDDR7. It remains to be seen whether its performance will be close enough to that of the former flagship to justify a price that, while lower, is still far out of reach for mainstream budgets.
A model of interest to many users could be the GeForce RTX 5070 Ti, which offers a balanced configuration with 16GB of GDDR7 memory and a high power consumption at 350W. This suggests a significant performance boost over the RTX 4070 Ti, which has a 285W power consumption. Of note, the RTX 4070 Ti is also available in a SUPER version with 16GB VRAM, and this model could see significant discounts in the near future.
Coming back to the RTX 5070 Ti, it will feature memory configured at an impressive 28 Gbps, giving a total bandwidth of 896 GB/s. This represents about a 33% increase over the bandwidth of the RTX 4070 Ti SUPER.
According to information from sources close to NVIDIA, the launch plan includes six RTX 50 models in the “Blackwell series,” which will be phased in during the first half of 2025. Initially, NVIDIA will introduce high-end models such as the RTX 5090 and RTX 5080, followed by the more affordable RTX 5070 Ti and RTX 5070 variants. The actual launch of these affordable models will largely depend on the strategy adopted by AMD, which is preparing its own range of GPUs, but also on Intel's response.
NVIDIA could make a last-minute adjustment to its pricing and performance strategy for the RTX 5070 if AMD's offering proves surprisingly competitive, in an attempt to maintain a market advantage.
