Investment Rating - The report maintains a "Buy" rating for the domestic intelligent computing chip sector, highlighting strong demand and advancements in performance and ecosystem [2]. Core Insights - The global demand for intelligent computing continues to surge, driven by large-scale AI model training and inference needs, with significant capital expenditures and supportive policies enhancing the market landscape [6][7]. - The domestic AI chip market is projected to grow at a CAGR of 53.7% from 2025 to 2029, with GPU market share expected to rise from 69.9% in 2024 to 77.3% by 2029 [18][20]. - The report emphasizes the importance of hardware-software synergy, showcasing advancements in chip performance and the development of independent software ecosystems to break the CUDA monopoly [6][7]. Summary by Sections 1. High Demand for Intelligent Computing - Global AI computing infrastructure investments are experiencing explosive growth, with major tech companies planning substantial investments in AI clusters, such as OpenAI's $500 billion "Star Gate" project [10][11]. - The daily token consumption in China has surged from 100 billion to 10 trillion within a year, indicating rapid adoption of generative AI across various sectors [13][15]. - Domestic capital expenditures in AI computing are being driven by major players like ByteDance, Alibaba, and Tencent, with significant investments planned for 2025 [23][24]. 2. Hardware Performance Breakthroughs - Domestic chip manufacturers are rapidly closing the performance gap with international competitors, particularly in advanced process nodes and single-card performance [6][7]. - Innovations in architecture, such as Huawei's CloudMatrix, demonstrate competitive capabilities against leading international solutions [6][7]. 3. Software Ecosystem Development - The report outlines the shift from compatibility adaptation to the establishment of independent standards in the software ecosystem, enabling domestic chips to compete effectively [6][7]. - Domestic companies are developing their own software stacks to reduce reliance on NVIDIA's CUDA, enhancing the overall ecosystem for AI applications [6][7]. 4. Investment Recommendations - The report suggests focusing on various segments within the intelligent computing industry, including chip manufacturers like Cambricon and Haiguang, server providers like Sugon and Inspur, and data center operators like GDS and Kuaishou [6][7].

Group 1: Industry Overview - The computational power required for large AI models has increased by 1000 times in just two years, significantly outpacing hardware iteration speeds [1] - Current AI training hardware is divided into two main camps: dedicated accelerators using wafer-level integration technology and traditional GPU clusters [1][2] Group 2: Wafer-Level Chips - Wafer-level chips are seen as a breakthrough, allowing for the integration of multiple dies on a single wafer, which enhances bandwidth and reduces latency [3][4] - The size of a single die chip is approximately 858 mm², and the maximum size is constrained by the exposure window [2][3] Group 3: Key Players - Cerebras has developed the WSE-3 wafer-level chip, which utilizes TSMC's 5nm process, featuring 4 trillion transistors and 900,000 AI cores [5][6] - Tesla's Dojo chip employs a different approach, integrating 25 proprietary D1 chips on a wafer, achieving 9 Petaflops of computing power [10][11] Group 4: Performance Comparison - WSE-3 can train models 10 times larger than GPT-4 and Gemini, with a peak performance of 125 PFLOPS [8][14] - In comparison, the WSE-3 has 880 times the on-chip memory capacity and 7000 times the memory bandwidth of the NVIDIA H100 [8][13] Group 5: Cost and Scalability - The cost of Tesla's Dojo system is estimated between $300 million to $500 million, while Cerebras WSE systems range from $2 million to $3 million [18][19] - NVIDIA GPUs, while cheaper initially, face long-term operational cost issues due to high energy consumption and performance bottlenecks [18][19] Group 6: Future Outlook - The wafer-level chip architecture is considered the highest integration density for computing nodes, indicating significant potential for future developments in AI training hardware [20]