The RTX 50 series has reportedly switched its GDDR7 memory chips from Samsung to SK Hynix.
GeForce RTX 5070 Reportedly Gets SK Hynix GDDR7 Memory but Other RTX 50 Series Have Adopted Memory Modules From SK Hynix as Well
NVIDIA has been outsourcing the memory chips from various memory manufacturers for its GPUs. Its main partners are SK Hynix, Micron, and Samsung, who provide the latest GDDR and HBM memories for NVIDIA's GeForce and data center GPUs.
As far as the latest RTX 50 series gaming GPUs are concerned, NVIDIA has been using the GDDR7 modules from Samsung, but the latest report suggests that the company has started accepting memory from SK Hynix as well. As reported by @Zed__Wang, NVIDIA started outsourcing the GDDR7 memory modules from SK Hynix recently, and its RTX 5070 is the first GPU to get these.
Nvidia recently started to use SK Hynix GDDR7 for the RTX50 Graphic Card. Started with RTX5070 first.
— MEGAsizeGPU (@Zed__Wang) April 8, 2025
The RTX 5090, RTX 5080, and RTX 5070 Ti have been utilizing Samsung GDDR7 memory modules since launch, but this report suggests that NVIDIA may have switched to SK Hynix for these GPUs as well. This means the next batches of these GPUs may feature SK Hynix GDDR7 memory. If you are concerned about whether this will affect performance, then the answer is no.
Typically, the performance and specifications will remain identical, and we will have 28 Gbps GDDR7 memory chips from SK Hynix as well. There might be some slight differences in memory overclocking and stuff, but that is generally not easier to find unless you are an enthusiast overclocker, who would want to tweak every parameter to leverage the full potential of the memory.
NVIDIA's plan to switch its memory partner right now coincides with SK Hynix's plans for mass production of GDDR7 memories. Samsung and Micron started GDDR7 mass production in late 2024 but SK Hynix started it very recently. In some previous reports, we heard that memory supply issues were affecting the availability of the RTX 50 series GPUs, and switching to SK Hynix might solve this problem.
It will be highly beneficial for NVIDIA to outsource memory from multiple partners rather than from a single one. This will enhance the availability of the RTX 50 GPUs drastically. At the moment, four of the RTX 50 series GPUs have been released and the RTX 5060 Ti and RTX 5060 are next in the queue. Both GPUs are expected to be released on 15th April but the RTX 5060 may see a delay in retail launch.
Follow Wccftech on Google to get more of our news coverage in your feeds.
