One of the "four domestic GPU dragons" Shi Yuan Technology's Science and Technology Innovation Board IPO has been accepted.

On January 22, Shanghai Suiyuan Technology Co., Ltd. (referred to as Suiyuan Technology), one of the "Four Little Dragons" of domestic GPUs, had its IPO on the Shanghai Stock Exchange Sci-Tech Innovation Board accepted. CITIC SEC is its sponsor institution, with a planned fundraising of 6 billion yuan. According to the prospectus, the company is one of the leading enterprises in China's cloud AI chip field, committed to becoming a leading enterprise in "general artificial intelligence infrastructure." The company adheres to the technological route of original innovation and independent research and development, building core competitiveness and moats for long-term sustainable development. Over the past nearly 8 years, the company has independently developed and iterated four generations of architecture for 5 cloud AI chips, establishing a complete product system covering AI chips, AI accelerator cards and modules, intelligent computing systems and clusters, and AI computing and programming software platforms. In the core technology areas, the company has accumulated over the years and formed three major categories: chip and hardware, software and programming platforms, and computing cluster solutions, creating a comprehensive and three-dimensional core technology system. In terms of underlying hardware, the company has based its architecture on its own instruction set, benchmarking NVIDIA's TensorCore acceleration computing unit and NVlink inter-card interconnect technology. The original GCU-CARE acceleration computing unit and GCU-LARE high-speed interconnect technology have not only the flexibility of programming but also deep support for AI large model high parallel computing. At the software platform level, the company has independently developed a full-stack AI computing and programming software platform "Yusan TopsRider" that includes drivers, compilation languages and compilers, operator libraries, and tool chains, rather than following NVIDIA's dominant CUDA ecosystem. This platform links the company's hardware with artificial intelligence applications, significantly reducing the programming development difficulty and migration cost of mainstream AI models based on the company's hardware, enabling the company's hardware products to better unleash performance in application scenarios. In terms of computing clusters, the company's smart computing center projects with a thousand cards and ten thousand cards have already generated revenue. Currently, the company has partnered with customers to develop super node solutions and jointly build high-value commercialized computing clusters with tens of thousands of cards. In terms of ecosystem construction, benefiting from years of deep cooperation with large Internet companies in software and hardware customization, the company's multiple generations of products have been widely commercialized in Internet AI scenarios, continuously providing AI computing power support for national-level Internet applications from traditional AI models to AI large models. The company's continuous iterative development capabilities and product competitiveness have been verified and recognized by the market, gradually achieving a key breakthrough from "technology product loop" to "commercial value loop." In addition, the company is fully leveraging its own advantages to jointly build an ecosystem with a wider range of downstream partners. In addition to participating in the national "East Digit, West Compute" hub node intelligent computing center project, the company is actively deepening cooperation with domestic network operators and exploring business opportunities in multiple vertical industries to provide inclusive computing power support for AI empowerment across various industries. Furthermore, to achieve sustainable development, Suiyuan Technology attaches great importance to long-term strategic cooperation with partners across the entire industry chain, including EDA/IP, wafer manufacturing, packaging testing, system components, etc., to ensure stable product development and supply delivery. The domestic cloud AI chip industry is in its early stages of development, with international manufacturers like NVIDIA holding a major share in the Chinese market. However, Chinese local enterprises are constantly breaking through technological barriers and gaining a certain market share. Chinese local cloud AI chip manufacturers include non-GPGPU architecture manufacturers represented by Huawei HiSilicon, Cambricon Technologies, and the company, and GPGPU architecture manufacturers represented by Mothread, Muxi, ILUVATAR COREX, and BIREN TECH. According to IDC data and company sales, the overall shipment scale of AI accelerator cards in China is expected to exceed 2.7 million by 2024, with NVIDIA's shipment volume of approximately 1.9 million accounting for about 70% of the market share. The company's AI accelerator card and module sales volume reached 38,800 units in that year, accounting for approximately 1.4% of the Chinese AI accelerator card market and ranking among the top domestic AI chip manufacturers. In the future, there is a certain degree of uncertainty in NVIDIA's sales in the domestic market, and the increasing market share of domestic AI chip manufacturers in the Chinese market is a high probability trend. The company's use of the funds raised is as follows: Financially, in the fiscal years 2022, 2023, 2024, and January-September 2025, the company achieved operating revenues of approximately 90.10 million yuan, 301 million yuan, 722 million yuan, and 540 million yuan, respectively; during the same period, the net losses were approximately 1.116 billion yuan, 1.665 billion yuan, 1.510 billion yuan, and 888 million yuan, respectively.