Workflow
并行科技(839493) - 投资者关系活动记录表
839493PARATERA(839493)2024-10-29 09:21

Group 1: Investor Relations Activity - The investor relations activity was conducted online on October 28, 2024, with participation from various securities and asset management firms [2] - The company representatives included the Chairman, General Manager, and Board Secretary [2] Group 2: Domestic Computing Chip Usage - Over 80% of computing resources can theoretically be replaced by domestic computing chips, but actual usage is significantly lower [3] - The company has successfully migrated a user to domestic computing chips, reducing costs to 80% of the original plan [3] Group 3: Client Base and Market Potential - The company serves over 300 research institutions, 400 universities, and 500 enterprises, indicating a broad client base [3] - Despite low current usage of domestic chips, the company anticipates accelerated adoption due to its extensive client network and technological advancements [3] Group 4: AI Cloud Computing Service Advantages - The company’s AI cloud computing services are built on IaaS and PaaS layers, providing tailored SaaS applications for various industries [4] - The company focuses on optimizing applications at multiple levels to enhance performance and user experience [4] Group 5: Resource Allocation and Revenue - The ratio of computing resources used for training versus inference is approximately 1:2, with training generating higher revenue [5] - The demand for inference computing is strong, while training demand from small to medium users remains high [6] Group 6: Collaboration and Market Expansion - The company collaborates with Zhiyuan, a key client, to provide AI computing and technical services [7] - Significant growth in the research user base in the new energy vehicle and intelligent driving sectors has been noted [8] Group 7: Cost and Profitability - The company utilizes both co-construction and external purchase models for computing resources, with co-construction yielding higher profit margins [9] - CPU resource utilization in co-construction mode exceeded 60%, while GPU utilization was around 80%-90% [11] Group 8: Project Implementation and Technical Advantages - The implementation of the "Supercomputing Cloud Computing Network Platform" project is progressing faster than planned due to increased GPU purchases [12] - The company’s core technical advantages include distributed supercomputing cluster operation, intelligent resource scheduling, and a leading SaaS service platform [13] Group 9: International Business Plans - The company has minimal overseas business income and does not own overseas assets [14]