Frost & Sullivan Releases the "2026 Global Novel Computing Power Chip and Future Key Technologies Development Industry Research Report"

In the global AI computing power chip competitive landscape, three major technology routes—GPGPU, DATAFLOW, and ASIC/DSA—each have their own advantages and are evolving in parallel. The general-purpose computing route represented by GPGPU has established a mature path in software ecosystem and large-scale deployment, while the novel computing power chip routes represented by DATAFLOW and ASIC/DSA are becoming a key force in addressing the computing power supply-demand gap, with outstanding performance in energy efficiency, computing efficiency, and flexibility.

The core characteristics demonstrated by novel AI computing power chips across five dimensions—high computing power, energy efficiency, flexibility, generality, and scalability—have helped the AI computing power chip industry break through three core pain points: weak generality, low computing efficiency, and insufficient scale expansion capability. Against the backdrop of computing demand shifting from training to inference, Moore's Law approaching its physical limits, and tightening external technology controls, companies need deep underlying architecture accumulation, a complete hardware-software product matrix, rapid iterative engineering capability, and a solid market position. By defining hardware through software, novel computing power chip companies are building differentiated competitive barriers and are becoming the new computing power foundation supporting the continued evolution of the AI industry.

Looking ahead, as cutting-edge technologies such as wafer-level integration, three-dimensional integration, and computing-network convergence continue to mature, the professionalism and innovation demonstrated by novel AI computing power chips will continue to drive the computing power industry toward greater efficiency and self-reliance, providing a replicable model for industry development and supporting sustainable growth.