Featured in PC Watch Column: LENZO’s CGLA Architecture in the NPU Era

In a recent semiconductor industry column by Yusuke Ohara for PC Watch, perspectives related to LENZO’s CGLA (CPU Grounded Linear Array) architecture were highlighted within a broader discussion of AI processors. The article explains why GPUs became central to AI processing, and why dedicated NPUs and AI accelerators are now gaining momentum.

The piece outlines how, as AI workloads continue to scale, conventional GPU architectures are beginning to show limitations in areas such as power efficiency, data movement, and sparse computation. In that context, dataflow-oriented designs and workload-specific accelerators are presented as increasingly important.

Within this broader shift, LENZO’s CGLA architecture can be understood as a new approach to structured compute - one designed to reduce unnecessary data movement and instruction overhead while improving efficiency for repeatable, structured workloads.

Read the full article here (Japanese Only)