Core Insights - The Smart Cloud Ecological Conference was held with the theme "Digital Intelligence National Cloud, Co-composing New Chapters" [1] - China Telecom's Tianyi Cloud Company introduced the latest technological achievement, the Triless architecture, during the conference [1] Group 1: AI Application Challenges - 2025 is identified as the year of AI application, with all future applications expected to be born, grow, and mature in AI [3] - Three major challenges for large-scale AI implementation were highlighted: efficient acquisition of computing resources, selection of training and inference frameworks, and convenient development of customized AI applications [3] Group 2: Triless Architecture Features - The Triless architecture aims to address the challenges by being "computing resource agnostic," "training and inference framework agnostic," and "AI tool agnostic" [3] - The "Xiangrang" platform upgraded its computing resource scheduling technology to connect diverse, cross-domain, and heterogeneous computing resources, allowing users to utilize computing power without being aware of underlying differences [3] - A self-developed heterogeneous training and inference framework was introduced, compatible with both domestic and NVIDIA computing resources, enhancing the performance of domestic computing and achieving compatibility with over 120 models [3] - A standardized AI development pipeline was launched, abstracting and encapsulating the AI toolchain into a ready-to-use development environment, allowing developers to focus less on tool differences [3]

Core Viewpoint - The conference highlighted the importance of accelerating the infrastructure of artificial intelligence (AI) technology through four key breakthroughs, as presented by the general manager of Tianyi Cloud, Hu Zhijiang [3]. Group 1: AI Infrastructure Breakthroughs - The first breakthrough is the standardization of AI capabilities, aiming to provide users with standardized AI service capabilities similar to communication services, without the need to understand underlying technologies [3]. - The second breakthrough focuses on the universal accessibility of AI services, which involves unified scheduling and efficient operation of computing resources to continuously lower computing costs [3]. - The third breakthrough emphasizes the popularization of AI applications, transitioning AI technology from a mere auxiliary tool for consumer (C-end) use to a productivity tool for business (B-end) applications [3]. - The fourth breakthrough is the self-sufficiency of the entire AI industry chain, recognizing AI as a critical battleground for major powers and advocating for the autonomy of key technologies and ecological reconstruction of the industry chain [3]. Group 2: Key Focus Areas for Tianyi Cloud - Tianyi Cloud is focusing on three key areas: platform architecture, domestic core technologies, and AI application ecology to achieve significant breakthroughs [4]. - The platform architecture has been enhanced with the introduction of the Triless architecture, which allows developers to focus solely on business logic without managing servers, thus lowering the barriers for large model application services [4]. - The domestic core technology initiative is moving towards integrated research and development of foundational software, ensuring self-sufficiency across the entire industry chain [4]. - The AI application ecology has been upgraded with the second generation of cloud computers and an AI application center, targeting various scenarios such as government, education, research, and management [4].