Floadia has successfully developed an innovative technology called CiM (Computing in Memory) that can reduce the power consumption of AI (Artificial Intelligence) semiconductors to about 1/1000 of conventional levels, and will begin offering evaluation chips from August 2025. This CiM technology achieves significant power reduction and space saving compared to conventional GPUs by processing AI calculations in an analog manner within non-volatile memory. In addition to edge AI applications such as drone control and autonomous driving assistance systems, we also plan to miniaturize AI data centers, which are facing major challenges due to increasing CO2 emissions and power consumption, from the space of a room to the size of a palm in the future, aiming to further improve the efficiency of AI calculations.

Details about this topics are published in the digital version of the Nikkan Kogyo Shimbun.
The article can be viewed in full (Japanese only) by registering as a free member. Please check it out.

Nikkan Kogyo Shimbun URL: https://www.nikkan.co.jp/articles/view/00753286