Although SK Hynix’s plans to begin manufacturing its HBM4 memory chips in 2025 were based on using the 5 nanometer manufacturing process, it finally seems that they will take advantage of the good evolution of the 3 nanometers from TSMC to launch production directly in this most advanced process of its high-end chips.
The main customer of these HBM4 memory chips will be NVIDIA, company that already would have pressured SK Hynix to advance its production and thus be able to implement them in its new accelerator GPUs during the second half of 2025. The rise of Artificial Intelligence has placed NVIDIA as the main supplier of accelerator cards, but the evolution is so rapid that the company seems to be in a hurry not to stay back and continue dominating the market with an iron fist.
The jump from 5 nanometers to 3 nanometers will mean an improvement in performance and efficiency that is expected to be around 40%.
The decision to use the TSMC process, a company with which they were already collaborating for the development of HBM4, will allow them to offer an advantage over the 4 nanometers of Samsung’s HBM4 memories. However, SK Hynix also has its own nodes and, in fact, conventional HBM4 memory chips They will be manufactured at 12 nanometers, leaving these 3 nanometers for very specific applications and clients.
End of Article. Tell us something in the Comments!
Antonio Delgado
Computer Engineer by training, editor and hardware analyst at Geeknetic since 2011. I love to tear apart everything that passes through my hands, especially the latest hardware that we receive here for reviews. In my free time I tinker with 3D printers, drones and other gadgets. For anything, here you have me.
Add Comment