Report: AI to Lead Edge Computing Market Growth "Cloud-Edge-End" Integration Architecture Development Expected

Wangsu Science & Technology, an edge computing service provider, recently released the "Edge Computing Market Practice and Insight Report" (hereinafter referred to as the "Report"), which proposed that edge computing has maturely applied in real-time audio and video interaction, CDN, cloud gaming, AR/VR, etc. In terms of edge cloud security, the integration of edge and security has become an industry consensus, and the SASE architecture has been widely deployed. In the IoT field, edge intelligent solutions have also achieved application innovations in industrial manufacturing, transportation, animal husbandry, smart cities, etc. IoT is one of the most demanding scenarios for edge computing, and the market will gradually form vertical industry edge intelligent solutions. At that time, the technical threshold and cost will be reduced, and the demand for edge computing will also grow accordingly.

According to the MarketsandMarkets report, the global edge computing market size will grow from US$60 billion in 2023 to US$110.6 billion in 2029, with a compound annual growth rate of 13%. According to the IDC report, the total size of China's edge cloud market in the second half of 2023 was RMB6.26 billion, up 36.1% YoY, exceeding expectations.

Chen Yunhui, senior architect of Wangsu Science & Technology's edge platform, said: "There are two major trends in AI development. On the one hand, large models deployed in the center will continue to develop towards greater intelligence and larger parameters. On the other hand, small models deployed on the terminal, such as the models being done by mobile phone manufacturers, need to be small enough in parameters while meeting certain performance requirements. We believe that AI large models are not only for the terminal and center, but also have opportunities in the edge."

Chen Yunhui said that for example, OpenAI's GPT-4o has added some multimodal and natural human-machine interaction experiences. Besides the improvement of model capabilities, the most important feature is real-time. The advantage of edge computing lies in low latency, which can play a role. Besides, the GPU computing power of mobile devices is limited, and the computing power needs to be overflowed for some scenarios. Compared with overflowing to the center, the delay may be 1-2 seconds, while overflowing to the edge side, which is closer to the user, the delay may be hundreds of milliseconds, which can meet the user's real-time interactive needs for generative AI, and the experience is more natural.

"So in general, AI has two needs for edge computing, one is computing power and the other is inference. Inference models need to be deployed on the edge side, and future inference and iteration will be distributed in a tiered manner in 'cloud-edge-end'. The integrated architecture of 'cloud-edge-end' is also a trend in the development of AI large models," Chen Yunhui said.