【純国産AI半導体】富士通が1.4nmチップをラピダスで量産する...推論市場の118倍需要とは

While NVIDIA dominates 92% of the AI ​​semiconductor market, Fujitsu has taken a completely different approach from GPUs. They plan to manufacture an inference-focused NPU using Rapidus's 1.4nm process, completing the entire process from design to mass production entirely domestically. What does this "different path" signify? VOICEVOX: Ryusei Aoyama *This video is for reporting and commentary purposes only and is not intended to criticize or defame any specific company or country. *This video was not created with the intention of infringing copyright. Quoted portions are used within the scope of reporting, criticism, and commentary. *This content is original content created independently and is not repeated or reused content. *While the information presented is based on meticulous research and analysis, its completeness is not guaranteed. We are not responsible for any damages incurred based on the information in this video. [Key Reference Information/Sources] NVIDIA Official Financial Results (FY2026 Q3) - AI Semiconductor Market Share and Sales Data Tom's Hardware - Fujitsu 1.4nm NPU Project Report Nikkei Asia - Details of Fujitsu NPU and Rapidus Collaboration Electronics Weekly - Details of NPU Development Costs and NEDO Support Rapidus Official Press Release - Funding and Factory Progress Bloomberg - Scale of Private Investment in Rapidus Japan Times - Rapidus 2nm Process Development Status IEEE Spectrum - Rapidus Technology Development Progress Fujitsu Official IR Materials - Management Structure and Semiconductor Strategy TrendForce - OpenAI's GPU Dissatisfaction and Inference Chip Market Trends SemiAnalysis / TechNode - Huawei Ascend Yield Problem Fujitsu Global / HPCwire - Fugaku Green500 Results METI / Digital Watch - Japanese Government's Semiconductor Investment George Glass, US Ambassador to Japan (X Post) - Rapidus Chitose Plant Evaluation