Analysis of AI Server Industry Development: Surging Demand for High-End AI Servers to Maintain Supply Shortage in Future Market
-
In 2023, China's artificial intelligence server market is projected to reach $9.1 billion, representing an 82.5% year-on-year growth. The intelligent computing power scale is expected to achieve 414.1 EFLOPS (exaflops per second), marking a 59.3% increase compared to the previous year. During the 2022-2027 period, the compound annual growth rate (CAGR) is forecasted to maintain 33.9%.
An AI server is a data server capable of delivering artificial intelligence (AI) capabilities. It can support both local applications and web services while also providing complex AI models and services for cloud and on-premises servers.
Since OpenAI officially launched ChatGPT in late 2022, its user base has grown exponentially, spawning numerous applications built around it. Its general-purpose capabilities have significantly reduced human effort in text-related tasks. Meanwhile, under the Transformer architecture, multimodal large models have achieved breakthroughs, with text-to-image and text-to-video functionalities continuously improving. These advancements have shown promising progress in fields like advertising and gaming.
According to OpenAI's documentation on the GPT-3 model, it contains 175 billion parameters and requires tens of millions of computational operations to complete a single inference task. ChatGPT's total computing power consumption is approximately 3,640 PF-days, requiring 7-8 data centers with an investment scale of $300 million each and individual computing power of 500P to sustain operations. Such scale and complexity demand high-performance computing equipment and massive parallel computing capabilities, driving the growth of AI data center requirements.
AI servers are the primary hardware for computing infrastructure, with GPUs accounting for the majority of training costs. Core components of AI servers include GPUs (graphics processing units), DRAM (dynamic random-access memory), SSDs (solid-state drives), RAID cards, CPUs (central processing units), network cards, PCBs, high-speed interconnect chips (on-board), and cooling modules.
The main CPU suppliers include Intel, while the leading GPU manufacturers are international giant NVIDIA and domestic companies such as Cambricon and Hygon Information.
Memory manufacturers primarily consist of Samsung, Micron, and SK Hynix, with domestic players like GigaDevice.
SSD manufacturers include Samsung, Micron, and SK Hynix, as well as domestic companies such as Longsys.
Overseas PCB manufacturers mainly include Gold Circuit Electronics, while domestic manufacturers comprise Wus Printed Circuit and Avary Holding.
Motherboard manufacturers include Foxconn Industrial Internet, and server brand manufacturers encompass Inspur Information, Unisplendour Corporation, Sugon, and ZTE Corporation.
Current Development Status of the AI Server Industry
The wave of the AI era is surging, with massive data generating enormous demand for computing power, driving the increasing demand for AI servers. Consequently, interface chips for data transmission inside and outside servers are also on the rise. In 2022, approximately 130,000 AI servers were shipped, accounting for about 1% of global server shipments.
By the end of 2022, among China's AI server manufacturers, Inspur Information ranked first in procurement volume with a market share of approximately 47%. H3C and Ningchang followed with procurement shares of 10% and 8% respectively, while other manufacturers collectively accounted for 35% of the market.
Currently, provinces across China are progressively increasing investments in AI-related fields. At the regional level, Guangdong, Jiangsu, and Beijing host the highest number of relevant enterprises. Statistical data shows that by the end of 2022, each of these regions had over 30,000 AI-related companies. With the continuous expansion of China's AI innovation application zones, multiple sectors represented by AI servers are gradually forming industrial clusters, leading to sustained improvement in regional concentration.
As artificial intelligence constitutes a crucial component of China's national development strategy, servers have expanded into various application domains within the current digital economy context. AI servers, as a rapidly developing emerging field, continue to experience market growth. Relevant national departments have successively introduced supportive policies for the industry. For instance, in February 2023, the Central Committee of the Communist Party of China and the State Council issued the "Overall Layout Plan for Digital China Construction," which proposed systematically optimizing computing infrastructure distribution, promoting efficient complementarity and coordinated development between eastern and western computing resources, and guiding rational hierarchical layouts of general data centers, supercomputing centers, intelligent computing centers, and edge data centers. These policy measures provide guarantees for the growing demand for AI servers.
Future Market Trends in the AI Server Industry
Driven by the wave of large AI models, the market size for AI chips, servers, and data centers is expected to grow significantly. IDC predicts that China's AI chip shipments will reach 1.335 million units in 2023, a year-on-year increase of 22.5%.
In terms of AI servers, IDC estimates that China's AI server market will reach $9.1 billion in 2023, an 82.5% year-on-year growth, and is projected to reach $13.4 billion by 2027, with a compound annual growth rate of 21.8%.
The surge in data required for large model training has created demand for high-end AI servers capable of supporting storage and training. Consequently, server manufacturers are upgrading chip specifications, expanding card configurations, and moving toward high-end AI server solutions.
With the acceleration of China's digital infrastructure development, the AI server industry is maintaining rapid growth. Major enterprises are making strategic deployments, and as AI application scenarios gradually materialize, the demand for computing power is growing rapidly. In 2022, China's intelligent computing power accounted for 22% of total computing capacity.
In 2022, general-purpose servers accounted for 93.2% of the total server market, while AI servers represented only 6.8%. As data servers that provide AI capabilities, AI servers are occupying an increasingly larger share of the overall server market. Future demand is expected to rise substantially. Industry experts predict that China's AI server market will remain in a state of supply shortage for an extended period.