Qualcomm is working on next-generation AI accelerators that integrate compute components beneath the DRAM. This architecture targets the memory bandwidth bottleneck that limits current AI hardware performance. By embedding compute under DRAM, Qualcomm aims to enhance AI infrastructure efficiency.
The company’s approach intends to help it catch up in the competitive AI hardware market. Overcoming memory constraints is a critical step for Qualcomm to improve performance and gain an edge in AI accelerator technology.
