One of the reasons NVIDIA’s high-end GPUs handle high resolutions so well, despite having a narrower memory bus than AMD, is because they use faster GDDR6X memory and larger caches. However, news has emerged that NVIDIA may be facing a shortage of GDDR6X modules. According to Board (via benchlife), the shortage could start in August.
GDDR6X memory is manufactured by Micron, but a quality issue with one of the batches may limit supplies. This may force NVIDIA to switch to slower GDDR6 memory for some models, such as the GeForce RTX 4070. However, such a change will require tweaks to the GPU architecture and changes to the software, which may impact performance, especially at higher resolutions, since the RTX 4070 has a 192-bit memory bus.
Supply issues won’t just affect the RTX 4070, but also higher-end models that use GDDR6X, like the RTX 4080/Super and RTX 4090. GDDR6X memory is the fastest available right now, with an effective speed of up to 23 Gbps on the RTX 4080 SUPER. By comparison, GDDR6 typically tops out at 18 Gbps, which can significantly reduce memory bandwidth.
However, this is just a rumor for now, and the supply problems may be temporary. NVIDIA may not have to switch to GDDR6, as the situation may stabilize. In this case, the company will most likely take measures to regulate supply and demand, and the Chinese market is expected to suffer the most.
Additionally, the RTX 40 shortage may not only be due to memory supply, but also to the growing demand for AI, which NVIDIA is currently focusing on. The company is also limiting the current generation’s production ahead of the expected launch of new gaming GPUs in early 2025. Either way, NVIDIA’s GeForce RTX 4070 and 4060 are likely to be hit the hardest. Whether the shortage is due to GDDR6X memory issues or AI demand, it will likely lead to higher prices for current-gen graphics cards.
