Recently, Japanese tech giant Fujitsu announced its participation in SoftBank Group’s AI memory development initiative. The project aims to develop next-generation high-performance, low-power memory technologies specifically optimized for artificial intelligence workloads, addressing the growing demands of AI models for faster data processing and improved energy efficiency. Leveraging its extensive expertise in high-performance computing, semiconductor design, and system integration, Fujitsu will collaborate with SoftBank and other partners to advance novel AI-optimized memory architectures.As generative AI and large language models continue to evolve rapidly, conventional memory technologies are increasingly struggling to meet the stringent requirements for bandwidth, latency, and power consumption during AI training and inference. SoftBank seeks to harness Japan’s domestic technological capabilities to build globally competitive AI infrastructure. Fujitsu’s involvement not only strengthens the project’s technical foundation but also reflects a strategic push by Japanese corporations to enhance autonomy and resilience in the domestic AI supply chain.The collaboration is also expected to explore cutting-edge approaches such as Processing-in-Memory (PIM), which could significantly boost the overall efficiency of AI chips. The resulting innovations may eventually be deployed in data centers, edge computing devices, and robotics, positioning Japan to play a more influential role in the global AI race.
近日，日本科技巨头富士通宣布加入软银集团主导的AI内存开发项目。该项目旨在研发专为人工智能工作负载优化的新型高性能、低功耗内存技术，以应对当前AI模型对数据处理速度和能效日益增长的需求。富士通将凭借其在高性能计算、半导体设计及系统集成方面的深厚积累，与软银及其他合作伙伴共同推进下一代AI专用内存架构的研发。随着生成式AI和大模型的迅猛发展，传统内存技术已难以满足训练和推理过程中对带宽、延迟和能耗的严苛要求。软银希望通过整合日本本土的技术资源，打造具有全球竞争力的AI基础设施。富士通的加入不仅强化了该项目的技术实力，也体现了日本企业在推动本土AI产业链自主可控方面的战略意图。据悉，该合作还将探索存算一体（Processing-in-Memory, PIM）等前沿架构，有望显著提升AI芯片的整体效率。未来，这一联合研发成果或将应用于数据中心、边缘计算设备乃至机器人等领域，为日本在全球AI竞赛中争取关键话语权。
原创文章，作者：admin，如若转载，请注明出处：https://avine.cn/5158.html