To meet the growing demand for computing resources driven by the spread of AI, we study advanced semiconductor technologies and develop our own high-efficiency AI chips.

Based on the latest research, we develop compilers and related system software that maximize the efficiency of computing hardware such as processors and chips.

We conduct research and development on high-speed, energy-efficient computing infrastructure that integrates our proprietary processors with cutting-edge hardware, software, and networking technologies.

Experts from diverse fields collaborate on research ranging from fundamental theory to industrial applications in machine learning. Our results are shared through publications, open-source software, and real-world systems.

To focus on what truly matters in solving real-world problems, we conduct research on automating and improving the efficiency of machine learning—centered around our open-source hyperparameter optimization framework, Optuna™.

We study cutting-edge foundation models and large language models that underpin generative AI, a technology increasingly essential to industry and society, and develop our own large language model, PLaMo™.

This research area focuses on enabling computers to “see”—to analyze and understand images and videos, and to reconstruct the real world in three dimensions. The technology has a wide range of applications, including robotics, autonomous driving, and the automation of factory and retail operations.

By combining deep learning with numerical computation, PFN develops materials simulation systems that achieve faster and more accurate predictions of material properties—accelerating materials discovery and development.

Using AI technologies, we conduct research and business development in fields such as omics analysis, medical image analysis, and drug discovery, aiming to apply these innovations to healthcare and life sciences.