For the past decade, Preferred Networks has been at the forefront of co-creating physical AI solutions with industry-leading companies in Japan, while designing a novel AI ASIC series focused on energy-efficient compute to power the wider, unprecedented growth of AI-based applications. In this talk, we will explore An AI-Semiconductor Nexus—a synergistic acceleration driven by specialized generative AI ASICs and AI-based material discovery—as a key driving force for future industrial scalability. We will first examine how Custom 3D-Stacked DRAM Architectures can deliver the desperately needed memory bandwidth to sustain the massive throughput required by the shift toward longer context in AI agent applications. We will then look at the transformative potential of Accelerated Materials Discovery, where AI-driven workflows are drastically compressing the R&D timeline. This ensures that innovations in the physical hardware foundation can keep pace with the continued exponential growth of AI and its industrial applications.

Thank you again for all your support and arrangements, and sorry for all your trouble.Very much looking forward to meeting everyone at the event tomorrow.