Nvidia GTC conference is about to be held to focus on the iteration of AI computing power

PANews March 18th news, according to financial reports, from March 17th to 21st local time in the United States, Nvidia will soon hold the top global AI summit - GTC2025. At that time, Huang Renxun will deliver a keynote speech to lock in the future development of AI agents, robotics and accelerate computing. This conference focuses on the iteration of AI computing power and will focus on displaying the new generation of Blackwell Ultra GPU and Vera Rubin super chip architecture. According to official disclosure, the conference will release a number of technological innovations, including the new generation of GB300 and B300 computing power cards, CPO switches and NVL288 cabinet solutions. Among them, the performance of B300 may be more than 50% higher than that of B200.

With the rapid development of AI technology, the demand for computing power continues to rise. In order to meet the growing computing power demand, hardware equipment is constantly iterating and upgrading, which has also brought huge business opportunities to the new materials field. Especially in the process of AI computing power iteration, the demand for high-performance and high-stability materials is becoming increasingly urgent, and PTFE (polytetrafluoroethylene) materials have become a core innovation point due to their advantages in high-frequency transmission performance. It is reported that in NVIDIA's new generation GB300 AI server and NVL72 architecture, PTFE-based multi-layer PCB (above 40 layers) is used in orthogonal backplane design, replacing traditional copper cable solutions. It is expected that the global demand for PTFE resin for AI servers will exceed 10,000 tons in 2025, and the market size will reach 10 billion US dollars.