Listing Success | Frost & Sullivan Supports Suzhou Denglin Technology Co., Ltd. in Its Successful Listing on Hong Kong Stock Exchange (9917.HK)

Over the past five years, as applications such as ChatGPT and other large model technologies have achieved breakthroughs, the global demand for AI computing power has surged, resulting in explosive growth in the GPU market size, reaching approximately RMB 1.7 trillion in 2025. Looking ahead, as Chinese large model technologies (such as DeepSeek), humanoid robots, and intelligent driving technologies advance, China’s level of AI technology will gradually reach international leading levels. Under these circumstances, the global economy is entering a new era of accelerated AI penetration, and the global GPU market size is expected to reach RMB 6.2 trillion by 2030. China’s GPU market has also experienced rapid growth, rising from RMB 64 billion in 2021 to RMB 185.3 billion in 2025. Given the ongoing diversification of AI applications and the strong demand for computing power infrastructure, it is expected that GPUs will become more widely available in China. By 2030, the Chinese GPU market size is projected to grow to approximately RMB 7,066 billion.

The general-purpose graphics processor (GPGPU) is a GPU product designed for general-purpose parallel computing tasks, suitable for artificial intelligence training, inference, and other high-throughput computing scenarios. Compared to dedicated accelerators designed for specific algorithms, GPGPU typically emphasizes parallel processing capabilities, programmability, and adaptability to diverse models and workloads. As AI models evolve rapidly, GPGPU has become an important component of AI computing chips.

From the perspective of downstream applications, China’s GPGPU market can be further divided into two sub-markets: public cloud AI and enterprise AI. Public cloud AI is primarily aimed at large cloud service providers and centralized computing infrastructure, while enterprise AI is targeted at enterprise private deployment and specialized AI computing scenarios across various industries. The two markets differ in deployment scale, procurement needs, performance requirements, and commercialization models, collectively driving the expansion of the GPGPU market.

• Dual-driven growth: AI training and inference – AI model iteration drives the growth in training computing power demands, while the commercialization of generative AI applications continues to expand the scale of inference computing, leading to demand for GPGPU products that possess generality and programmability.

  
  

• Diversified computing needs: With the development of industry models, visual algorithms, and AI smart agents, AI workloads become increasingly diverse in terms of model architecture, input types, computational accuracy, and deployment environment, posing higher requirements for the general adaptability of GPGPU.

  
  

• Localized ecosystem and product adaptation: Corporate clients are increasingly focusing on computing efficiency, software framework compatibility, deployment costs, supply chain stability, and local technical services, prompting domestic GPGPU enterprises to establish differentiated capabilities around synergistic software and hardware and application adaptation.

• Technical barriers:The GPGPU industry possesses high technical entry barriers. High-precision floating-point operations, as one of the core technologies in the industry, requires enterprises to have deep technical expertise across multiple aspects such as chip architecture design, circuit optimization, and algorithm coordination. As enterprise AI computing and other application scenarios continue to expand, GPGPU products must not only provide robust computing power output but also continuously optimize energy efficiency, involving advanced manufacturing process selection, power management module design, and energy efficiency optimization of computing units—all of which involve significant challenges, making it difficult for new entrants to overcome technical barriers in the short term.

  
  

• Supply chain barriers: The GPGPU industry also has high supply chain entry barriers. At the upstream, GPGPU design companies must establish long-term, stable cooperative relationships with wafer foundries, accurately monitor capacity and process capabilities, and ensure a stable supply of high-quality wafers through supply chain coordination. At the downstream, companies must meet the multi-dimensional requirements of their core customers regarding product performance, quality, and delivery, as well as the ability to quickly respond to changes in customer needs. Additionally, the integration of upstream and downstream industrial chain resources and supply chain coordination management require long-term professional knowledge and practical experience, further increasing the entry barrier for new entrants.