The U.S. ban hammer currently restricts CXMT from getting its hands on the HBM manufacturing tools needed to stay competitive in mainstream AI memory. However, the Chinese memory maker may have found its true calling in the form of customized DRAM. By diving headfirst into 3D memory and 3D Integrated Circuits (3D ICs), a fresh report highlights a strategy CXMT could employ to forge its own path and potentially catch South Korean titans like Samsung and SK Hynix off guard.
As AI hardware moves beyond GPUs and towards high-speed memory, CXMT wants a massive profit chunk
Being limited to existing manufacturing equipment has limited CXMT’s progression towards the HBM market, but it also presents a major opportunity for China’s memory makers to carve out their own path. More specifically, ZDNet reports that there’s a strategic shift towards 3D Integrated Circuits (3D ICs) and 3D DRAM that could bypass performance and data bottlenecks without relying on HBM.
Fabless semiconductor manufacturers have reportedly inquired Samsung about these designs, but the Korean giant and SK hynix are proceeding cautiously as they don’t want their current cash cow to be compromised. Additionally, Samsung and SK hynix make the bulk of their profits through high-volume shipments of standard DRAM modules.
To shift to a completely niche and customized DRAM market goes against their business model while also introducing major risks that these manufacturers would completely avoid. Even though these memory titans are said to be conducting advanced research on 3D DRAM and 3D ICs, they won’t be running wild on commercialization anytime soon. The safe approach adopted by these firms gives CXMT a major opportunity.
However, we haven’t addressed the challenges that come with tackling a customized and niche market. For instance, despite China’s semiconductor firms having developed multiple 3D DRAM sample chips and running active customer projects, uncertainty looms over this architecture’s ability to equal or surpass HBM in AI inference performance. The cost is another problem, as CXMT might gain the advantage over Samsung and SK hynix, but will the market be worth it?
There’s no information about how well the yields will hold up when the first batch enters volume production, meaning that a significant dollar amount will be wasted in putting a certain technology through the trial-and-error process. Again, success isn’t guaranteed, but it’s definitely better than waiting for HBM equipment to become accessible to Chinese firms, which they won’t.
News Source: ZDNet
About the author: Omar Sohail is a reporter and analyst for Wccftech's mobile section, specializing in the technology and business of the mobile industry. His expertise lies in the intricate hardware supply chain, covering developments in semiconductor manufacturing, chip lithography, and camera sensor technology.
Follow Wccftech on Google to get more of our news coverage in your feeds.
