On December 11th, Gelonghui reported that Marvell Technology announced on Wednesday that the company has made breakthroughs in customized high-bandwidth memory (HBM) for AI processors. It is reported that the new AI accelerator architecture (XPU) will increase computing power by 25%, memory by 33%, while also improving energy efficiency. The company is working with Micron Technology, Samsung, and SK Hynix to customize HBM processors for the accelerator. Marvell Technology executive Will Chu stated in a statement, "Enhancing XPU by customizing HBM for specific performance, power consumption, and total cost of ownership is the latest step in the new paradigm of AI accelerator design and delivery. We are very grateful to collaborate with leading memory designers to accelerate this revolution and help cloud datacenter operators continue to expand their XPU and infrastructure for the AI era."