Core Insights - The 2025 Open Computing Technology Conference held in Beijing focuses on the development trends of MoE large models and AI agents, emphasizing the importance of open computing in enhancing both vertical scaling and horizontal efficiency [1] - Open-source large models are reshaping the global AI industry landscape, significantly lowering the barriers for enterprises and individual developers to access advanced AI capabilities, thus accelerating the shift from closed to open collaboration [3] - The rise of open computing is fostering tighter collaboration within the data center industry chain, which is crucial for the rapidly evolving AI sector [4] Industry Developments - The MoE large models are experiencing rapid growth in parameter counts, necessitating innovations in computing architecture to meet the extreme demands for computing density and interconnect speed [4] - The power requirements for AI data centers are projected to escalate from over 100 kW per cabinet to above 1 MW, indicating a shift towards GW-level power demands [4] - China has a significant advantage in energy infrastructure, particularly in renewable energy, with 90% of new installations in Q1 2025 coming from renewable sources, contributing to 35.9% of the total power generation [5] Collaborative Efforts - The establishment of the "GW-level Open Intelligent Computing Center OCP China Community Group" aims to leverage China's strengths in energy and computing infrastructure to promote the implementation of AI open system strategies [5] - OCP is actively collaborating with OCTC to explore the deployment of advanced AI infrastructure technologies and research outcomes in the Chinese market [5] - Future initiatives will focus on creating a global open-source coalition to enhance collaboration among developers across different countries and regions, promoting innovation and integration within the global supply chain [6]