天地一体协同计算
Search documents
航天宏图(688066.SH):自主研发的PIE-Engine时空云计算平台是公司空间计算的基础平台
Ge Long Hui· 2025-09-18 07:48
Core Viewpoint - The company is focusing on the development of space computing capabilities, transitioning from traditional data processing methods to in-orbit data processing, which enhances efficiency and application in various fields [1] Group 1: Space Computing Infrastructure - Space computing refers to deploying data processing facilities, such as servers and AI chips, in space to handle massive data in orbit, moving from "ground processing" to "in-space processing" [1] - The company's self-developed PIE-Engine cloud computing platform serves as the foundational platform for space computing, providing data and computing power [1] Group 2: Technological Advancements - The PIE-Engine cloud platform integrates with the DeepSeek model, achieving a disaster recognition accuracy rate of 92%, showcasing strong data processing and analytical capabilities [1] - The VR/XR team leverages its experience in virtual reality to support future space computing initiatives alongside the three-dimensional platform [1] Group 3: Strategic Framework - The company has established a three-dimensional strategy combining satellite operations, low-altitude economy, and data elements, creating a closed loop of "data collection - processing - application" [1] - Through the "Nüwa Constellation," the company collects data, processes it on its platform, and applies it across multiple sectors, positioning itself to benefit from the growth of the space computing industry [1]
太空算力:打破传统地天观,重构天基智能版图
Yin He Zheng Quan· 2025-07-14 09:27
Investment Rating - The report maintains a "Recommended" rating for the communication industry [3] Core Insights - Space-based computing is transforming the traditional "ground-based computing" model into a new paradigm where data processing occurs in orbit, significantly enhancing efficiency and reducing data wastage [8][19] - The market for space computing is vast, with applications expanding from remote sensing to real-time scenarios like vehicle networking and drone clusters, potentially opening up a trillion-dollar market [5][19] - The report highlights the advantages of space computing, including energy efficiency, superior thermal management, and modular deployment, which collectively address the challenges of energy consumption, density, and scalability in AI infrastructure [19][25][26] Summary by Sections 1. Breakthrough in Space Infrastructure - Space-based computing allows satellites to evolve from mere sensing platforms to intelligent entities capable of data processing and storage [8] - Traditional models waste significant amounts of data due to bandwidth limitations, with only about 4% of data from solar monitoring satellites being utilized effectively [8][11] - The transition to "space-based computing" enhances real-time task capabilities and has broad strategic value [8][12] 2. Domestic and International Case Studies - Starcloud, a U.S. startup, plans to launch the first AI satellite equipped with NVIDIA H100 by 2025, aiming to create a gigawatt-scale orbital data center [5][39] - The "Three-Body Computing Constellation" in China, launched in May 2025, features 12 intelligent satellites with a total computing power of 5 POPS, marking a significant advancement in space computing capabilities [5][44][49] 3. Outlook for Ground Communication Operators - Putian Technology is transitioning to become a provider of space-based information infrastructure, focusing on low-orbit satellite communication and high-reliability satellite manufacturing [5][30] - The report suggests that space computing can provide new perspectives for industries like finance and disaster monitoring, with significant market potential [5][19] 4. Investment Recommendations - The report recommends focusing on satellite ground operators like Putian Technology, which show significant marginal improvements [5][19]