經過二十年的發展，雲計算已成為數字經濟的主要服務形態，有效支撐了各行各業數字化轉型。但隨著千億參數大模型等AI技術的迅猛發展，傳統雲服務在算力供給、網路傳輸及服務形態等方面已難以滿足需求，迫切需要構建適應智慧時代的雲計算新範式。4月10日，中國移動發佈《雲智算技術白皮書》，業界首次提出下一代雲計算範式——雲智算技術體系和十大關鍵技術方向，推動“雲計算”向“雲智算”升級，帶領雲計算產業技術革新。
At the important moment of the transition of technology and industry from cloud to intelligence, China Mobile took the lead in proposing a cloud intelligent computing paradigm with AI as the core driving force, and upgraded the cloud intelligent computing architecture from the three layers of IaaS (infrastructure as a service), PaaS (platform as a service) and SaaS (software as a service) of traditional cloud computing to four layers of AI IaaS, AI PaaS, MaaS (model as a service) and AI SaaS, promoting the transition from "computing network brain" to "computing network intelligent brain". Among them, AI IaaS is an infrastructure service provided for the integration of computing and networking, which promotes the "joint computing network" of multiple types of computing power through the ubiquitous network, and relies on the computing network brain to realize flexible scheduling and out-of-the-box use of computing power. AI PaaS is a tool platform service for all kinds of AI developers, providing a tool chain and development environment covering the entire process of AI R&D, operation, and testing, significantly improving the efficiency of AI innovation in the whole society. MaaS is a model service that accelerates the one-stop implementation of AI, bringing together resources such as models, capabilities, and agents to promote the popularization and penetration of AI in various industries. AI SaaS is an AI application service covering diverse scenarios, enabling the digital and intelligent transformation of production methods, lifestyles, and social governance methods. Through the deep integration of computing network infrastructure and AI core technology, China Mobile can provide a new cloud service model of integrated computing network resources, full-stack development environment, one-stop model services, and diversified scenario applications, effectively reducing the computing cost of enterprises and individual users, and realizing the convenient development of AI applications.
The "Cloud Intelligent Computing Technology White Paper" released this time systematically defines the cloud intelligent computing architecture and proposes ten key technical directions, covering cutting-edge fields such as computing power chips and large model services, providing a technical roadmap for the development of cloud intelligent computing in the industry. (Reporter Li Zhengwei)
Source: Bright.com