Xiao Qiyang was born in Hong Kong, China, and earned his Ph.D. from Stanford University at just 24 years old. Marvin Minsky, the father of artificial intelligence, even wrote the preface for his book. It wasn't until 2019 that Xiao Qiyang prepared to start a business, aiming to tackle the challenge of developing DPU (Data Processing Unit) chips. He believes that in cloud services, all the infrastructure tasks that CPUs handle poorly and GPUs cannot handle can be delegated to the DPU. The DPU can be considered the very soul of the entire cloud data center. Since the end of last year, CPU companies represented by Haiguang and Loongson, along with GPU or AI computing power chip companies like Cambricon, Moore Threads, MetaX, and Biren, have all entered a harvest season in the secondary market. The DPU company founded by Xiao Qiyang, Shenzhen-based Yunbao Intelligent (Cloud Leopard), has also entered the capital market's view. Yunbao Intelligent's filing for listing tutoring was recently accepted by the Shenzhen Securities Regulatory Bureau, with CITIC Securities as the tutoring institution, marking its journey to become "China's first DPU stock." This is the first industry unicorn in the domestic DPU sector, with a valuation of 14 billion yuan after its last round of financing. In the same year the company was founded, NVIDIA founder Jensen Huang also publicly stated that the "DPU is the third major chip for data centers," leading to the DPU's astonishingly rapid development and becoming a key focus for the entire industry. With the acceleration of artificial intelligence and the construction of hyperscale AI computing centers, the importance of DPUs is becoming increasingly prominent. Currently, major chip manufacturers like NVIDIA, Intel, Broadcom, and Marvell have already laid out their strategies in the DPU arena. Facing a competitive landscape where international tech giants dominate the global DPU market, domestic manufacturers have vast space for independent innovation. Companies like Yunbao Intelligent are charting a different path by focusing on creating domestic DPUs with the "best cost-performance ratio." Tencent Bets, Becomes Largest Shareholder Yunbao Intelligent's founder, Xiao Qiyang, was born in Hong Kong, China, and studied in the United States early on, earning a Ph.D. in Electronic Engineering from Stanford University at age 24. During his studies, his doctoral dissertation solved a theoretical problem in the field of artificial intelligence that had remained unsolved for over 30 years, prompting Marvin Minsky to write the preface for his work. After graduation, Xiao Qiyang first taught at UC Irvine, then served as an endowed chair associate professor at MIT, where he founded the Automatic Identification Research Center. His early research focused on AI neural networks, later shifting to high-performance processors and network chip-related studies. In 2002, Xiao Qiyang left academia and co-founded RMI Corporation in Silicon Valley. Reports indicate that at RMI, Xiao and his team developed the world's highest-performance multi-core processor for the communications sector from scratch. Its chips were widely adopted by several domestic and international communication network equipment manufacturers in routers, wireless base stations, controllers, core networks, security devices, and smart network interface cards. In 2009, RMI merged with NetLogic Microsystems. Two years later, Broadcom acquired NetLogic for $3.7 billion. Five years after that, Xiao Qiyang left Broadcom. A leading domestic cloud service provider was preparing to develop smart network interface cards and hired seven core members of his team. In terms of chip form, the DPU evolved from the smart network interface card. The single-chip SoC technical solution is considered the future mainstream for DPUs due to its advantages in high performance, low power consumption, and general programmability, although its technical barriers are several times higher. In 2019, driven by the rapid development of cloud computing, data centers, and AI, alongside changes in the international tech environment, Xiao Qiyang—already financially independent and possessing a deep understanding of smart NICs—prepared to start his own business to conquer DPU chips. Xiao Qiyang has stated, "In cloud services, all the infrastructure tasks that CPUs handle poorly and GPUs cannot handle can be delegated to the DPU. The DPU can be considered the very soul of the entire cloud data center." In August 2020, Xiao Qiyang established Yunbao Intelligent, a leading semiconductor company focused on cloud computing and data center DPU chips and solutions. According to its website, the core team hails from Broadcom, Intel, ARM, Alibaba, and others, possessing extensive experience in DPU chip and software R&D. In the very same year Yunbao Intelligent was founded, the DPU赛道 received major backing from giants. Jensen Huang stated in 2020 that the DPU would become one of the three pillars of future computing, with the future data center standard being CPU + DPU + GPU: CPUs for general-purpose computing, GPUs for accelerated computing, and DPUs for data processing. Concurrently, NVIDIA acquired the global smart NIC leader Mellanox for $6.9 billion, making a significant bet on the DPU赛道. As the DPU赛道 gained more attention, major internet companies and market investment institutions joined the fray. Shortly after its establishment, Yunbao Intelligent secured seed-round funding from institutions including Yaotu Capital, Hongzhuo Capital, 5Y Capital, Zhenxingu Capital, China IC Capital, and Huaye Tiancheng. Subsequently, Yunbao Intelligent continuously secured capital injections. By its Series D funding round at the end of November last year, its valuation had exceeded 14 billion yuan, with participants including Tongchuang Weiye, Oriental Fortune Capital, Cornerstone Capital, IDG Capital, and Tencent. Regarding equity structure, Qichacha information shows that through multiple investments, Tencent holds a combined stake of 22.5351%, making it the largest shareholder. Founder Xiao Qiyang directly holds a 7.551% stake. Building the "Best Cost-Performance" Domestic DPU Yunbao Intelligent has become the sole unicorn enterprise in the domestic DPU chip field, continuously attracting capital investment, which is underpinned by the company's technology commercialization and business applications. Public information indicates that in the global DPU technology race, Yunbao Intelligent independently developed China's first high-performance, generally programmable DPU SoC, which currently reaches top industry standards. Compared to traditional solutions, the Yunbao DPU chip solution offers a 4x performance improvement, reduces power consumption by over 50%, achieves network bandwidth of 400 Gbps and RDMA at 200 Gbps, leading to a significant cost reduction. This chip employs an innovative hierarchical programmable design, integrating a data processing unit supporting the P4 language, a self-developed RISC-V microprocessor unit, and a high-performance server CPU processing unit. It aims to comprehensively cover diverse user needs and business customization, significantly enhancing data center computational efficiency. Currently, this Yunbao chip has entered mass production and is widely used by leading domestic cloud service providers and operators like Tencent and China Mobile, with applications across multiple industries. Specifically, the Yunbao Yunsiao DPU chip is China's first DPU chip to reach 400Gbps. It is applied in scenarios like cloud computing and artificial intelligence, achieving triple benefits in data center performance, energy efficiency, and security. Additionally, the Yunbao Fengchi smart NIC chip is a leading domestic 400Gbps smart NIC chip. It serves as the core chip supporting the interconnection of tens of thousands of GPUs, significantly improving the effective computational utilization rate of clusters, and is a key infrastructure chip for training trillion-parameter models. Xiao Qiyang stated that Yunbao Intelligent's goal is not merely to pursue singular "peak performance" but to create domestic DPUs with the "best cost-performance ratio." This approach aims to meet the objective of rapid product commercialization and capture the largest market share. Furthermore, it utilizes the core chip architecture to support high-performance, high-throughput virtualization and various cloud and AI applications, thereby enhancing operational efficiency and reducing hardware and operational costs. Amid the current global wave of artificial intelligence, the DPU, as a new type of processor for data-centric computing, forms the three core components of modern computing alongside the CPU and GPU. It is primarily applied in data centers for security acceleration, storage acceleration, and network virtualization, acting as the data center's "intelligent traffic and logistics hub." If a computing center is likened to a modern city, the CPU is the city's brain, responsible for overall decision-making and diverse complex logic. The GPU is the specialized muscle group, tackling tasks like AI computing and graphics rendering. The DPU, then, is the city's intelligent traffic and logistics hub, efficiently dispatching, distributing, and securing all data flows to ensure resource coordination. Today, several world tech giants, including NVIDIA and Intel, along with major domestic cloud service providers like Alibaba Cloud, Tencent Cloud, Baidu Cloud, and Volcano Engine, almost universally use DPUs in their respective data centers for elastic scheduling of computing resources, dynamic on-demand allocation, and more, aiming to enhance the efficiency of cloud service computing resources and reduce operational costs. The surge in AI-driven demand for data center computing power positions the DPU as a critical chip for accelerating data processing, ushering in a period of rapid expansion for the global DPU market. While international giants like NVIDIA, Broadcom, and Intel dominate this market, the intensely competitive landscape creates broad space for independent innovation by domestic manufacturers. It is understood that Yunbao Intelligent plans to further expand its industrial chain in the future, establishing a strategic investment department to participate in investing in upstream and downstream companies within the sector, thereby extending its business development and creating synergies.

