Driven by the surge in AI technology, the cloud computing industry is undergoing unprecedented transformation. With the surge in the number of parameters in large models, traditional cloud computing architecture is shifting towards an “intelligent engine” paradigm. This shift not only tests cloud vendors’ ability to cope with the demand for 100x more computing power, but also profoundly impacts the logic behind enterprise cloud migration and capital market pricing mechanisms.
Recently, the “Hai Tan Technology” cloud computing forum, a special discussion focused on the latest trends and challenges in cloud computing, was successfully held, attracting numerous industry experts. Professor Qi Zhengwei of the School of Computer Science at Shanghai Jiao Tong University, Tang Wenjun, Senior Solutions Architect at Beijing QingCloud Technology, and Yang Lin, Chief Computer Analyst at Cathay Haitong Securities, engaged in an in-depth discussion on the development paths and investment opportunities of cloud computing in the AI era.
The rapid development of AI technology is having a profound impact on the cloud computing sector. On the one hand, computing power demand is expanding both vertically and horizontally, driving the continued expansion of the cloud computing industry. Tang Wenjun pointed out that traditional CPU-dominated cloud computing architectures are no longer able to meet the parallel computing requirements of large AI models, and cloud vendors need to upgrade their computing power reserves to a model primarily based on accelerator cards such as GPUs. At the same time, inference scenarios for large AI models are gradually spreading to the edge, prompting the gradual formation of a computing system that collaborates across cloud, edge, and edge.
On the other hand, the “horizontal diffusion” of computing power from the cloud to the edge and the “vertical expansion” of rack capacity are in tandem, shaping the future of cloud computing. Professor Qi Zhengwei believes that future data centers will achieve rack-integrated computing, storage, and communication networks. Intelligent computing, supercomputing, and general-purpose computing power will be aggregated in the same resource pool, forming a scalable, large-scale computing system.
AI technology has not only driven technological innovation in cloud computing but also driven rapid market growth. Data shows that the global cloud computing industry is expected to grow from $700 billion in 2024 to nearly $2 trillion in 2030, while my country’s cloud computing market is expected to exceed 3 trillion yuan by 2030, up from 800 billion yuan in 2024.
From an upstream and downstream perspective, the widespread adoption of AI technology is triggering a series of chain reactions. Analyst Yang Lin pointed out that while the computing power required for model training still accounts for a high proportion of the industry, as models mature and application scenarios expand, the demand for inference computing power, especially on the edge and edge, will continue to rise. For vertical application software vendors, SaaS products that combine AI capabilities will usher in an explosion of market demand.
The booming cloud computing market has created unprecedented market opportunities for both software and hardware technologies. Since the release of GPT-3.5, the stock prices of AI accelerator card manufacturers such as NVIDIA have soared, and the computing power rental industry has subsequently entered the capital market’s radar. As the influence of AI continues to expand, hardware such as HBM, optical modules, and liquid cooling are also gaining market favor. Yang Lin predicts that China’s AI applications will experience a surge in the second half of this year or next year, benefiting computing power chip manufacturers, optical module manufacturers, server manufacturers, computing power rental companies, and IDC companies. At the same time, market opportunities will also emerge for software and middleware layers that complement the computing power hardware layer. Professor Qi Zhengwei emphasized that the integration of software and hardware will be a key focus of future technological development. Tang Wenjun further pointed out that the trend of “hardware-software integration” will primarily manifest in optimizing the software itself, optimizing software around hardware, and optimizing the middleware between software and hardware. For example, Huawei is gradually open-sourcing the middleware for its Ascend computing power cards to attract more developers to build a collaborative ecosystem and promote the efficient operation of various programs on the accelerator cards.
Looking ahead, technical fields such as AI programming and AI marketing are expected to be the first to achieve commercialization. These fields boast abundant high-quality data, advanced technology, relatively closed scenarios, and high fault tolerance, creating favorable conditions for the commercial application of AI technology. For example, in the field of AI programming, relatively mature products such as Cursor have emerged and secured significant funding. AI-powered creative content generation, such as video content, has become widespread in the marketing industry, and some business processes in industries like finance are also being reshaped by AI.
