Fujitsu has announced that its comprehensive traffic simulation system has been adopted for Maebashi City's Regional Public Transportation Plan, published on 23 March 2026. The system, developed under Japan's Ministry of Land, Infrastructure, Transport and Tourism-led COMmmmONS project, is the first in Japan to simulate both fixed-route and demand-responsive transportation. The technology uses Fujitsu's social digital twin capabilities to model human and social behaviour, drawing on statistical data about residents and ridership information from MaaS applications. Maebashi City used the system to provide scientific evidence supporting bus route expansion measures in its public transportation strategy. Fujitsu plans to commercialise the system by fiscal year 2026, positioning it as a standard tool for local governments across Japan addressing transportation challenges and carbon neutrality goals.

Arrcus has announced a collaboration with Fujitsu and 1Finity around FUJITSU-MONAKA, Fujitsu's next-generation Arm-based CPU, to deliver secure, energy-efficient infrastructure for distributed AI workloads. The partnership aims to address challenges in latency, power consumption, data sovereignty and security as AI inference expands beyond centralised data centres to the edge. The architecture integrates FUJITSU-MONAKA-powered compute with Arrcus ArcOS network operating system and 1Finity's optical interconnect technology. It is designed to support use cases including physical AI in smart factories and robotics, enterprise AI workloads requiring localised processing, and service provider-delivered AI services. FUJITSU-MONAKA features built-in confidential computing capabilities and improved power efficiency. The collaboration was showcased at MWC Barcelona, where both companies demonstrated the joint solution.

Broadcom is shipping a new custom AI chip design to Fujitsu that uses advanced stacking technology to improve energy efficiency. The company expects major data centre operators to adopt the design later this year. The 3.5D eXtreme Dimension System in Package stacks two chip components top to top, rather than the top-to-bottom approach pioneered by AMD. This configuration enables greater data transfer with better power efficiency, according to Harish Bharadwaj, vice president of marketing in Broadcom's custom chip unit. The development represents Broadcom's push into the custom AI chip market, targeting data centre operators seeking more energy-efficient solutions for AI workloads.