ICs

In recent years, amid the need for the efficient operation of equipment and machinery, the application of AI to predict and detect faults and improve maintenance efficiency have emerged as key themes. On the other hand, conventional AI processing models necessitate network connectivity, high-performance CPUs, and so forth, which posed challenges with cost and installation environments. To address such challenges, as of March 2025, ROHM has developed the industry’s first^*1 microcontroller equipped with AI functionality, which makes both learning and inference possible on the microcontroller itself without requiring a network. This product utilizes a proprietary AI accelerator^*2 to achieve AI processing speeds 1,000 times faster than ROHM’s microcontrollers employing conventional software-based systems (Condition: theoretical value at 12 MHz operation). It is also able to flexibly accommodate disparities across installation environments and devices. Additionally, ROHM offers a simulation tool online, called “Solist-AI™ Sim,” that makes it possible to verify the product’s effects before implementing it. We also have a model development and implementation support structure in place through an ecosystem in collaboration with partner companies. In this way, by detecting signs of device failure before it occurs and lowering both maintenance costs and the risk of line stoppage with its microcontrollers equipped with AI functionality, ROHM contributes to the stable operation of equipment and devices and to the further enhancement of their efficiency.