Group 1 - The global AI competition is shifting from a "sprint" of technological breakthroughs to a "marathon" of ecosystem building, with key variables being computing power costs, data quality, and scene implementation capabilities [1] - China aims to leverage its advantages in "green electricity + domestic chips," "data flow," and "scene sinking" to create a unique development path that combines technological breakthroughs with social benefits [1][2] - The high energy costs and chip supply limitations are major constraints on the global AI industry, and China's solution lies in integrating its renewable energy advantages with independent innovation to build a low-cost, high-security computing power supply system [1][2] Group 2 - The "Westward Migration" strategy of data centers is reshaping the cost structure of computing power, with regions like Qinghai and Inner Mongolia offering significantly lower electricity prices, thus reducing the costs of large model training [1] - The innovative "modular" chip cluster solution developed by Chinese companies allows for the combination of domestic 14nm chips to achieve performance equivalent to 3nm chips, drastically reducing training costs from tens of millions to millions [2] - The challenge of "data islands" and security concerns hinders the transformation of vast data into innovative momentum, but China's large-scale data base can provide continuous "live water" for AI development if data circulation barriers are broken [2][3] Group 3 - Innovations in data circulation are emerging across the country, with examples like Shenzhen's data element market and Hangzhou's "city brain" demonstrating the potential for increased efficiency when data is treated as a public resource [3] - The ultimate value of AI lies in solving real-world problems, and China's unique advantage is its comprehensive application scenarios from urban to rural areas, enabling AI to evolve from a "toy" to a "tool" [3][4] - Scene sinking is transforming traditional production methods, with AI applications in community markets and agriculture enhancing productivity and making technology accessible to ordinary people [4] Group 4 - The successful implementation of these three strategies relies on precise policy guidance, balancing innovation with resource management to prevent waste while allowing for experimentation [4]

Core Viewpoint - The 2025 Global Digital Economy Conference highlighted the importance of high-quality data set construction and the establishment of a trusted data space as key focus areas for future data circulation [1][3]. Group 1: Data Circulation Infrastructure - Beijing Capital Information Development Co., Ltd. is leading the construction of a "Data Circulation Utilization Value-Added Collaboration Network" to address the supply-demand mismatch of high-quality data sets [3]. - The data circulation platform has officially launched, offering nearly 4,000 data products across 12 application scenarios and aggregating over 300 market entities [3]. - The company aims to enhance the market by increasing the availability of high-quality data sets on the platform, facilitating the flow and transaction of data between supply and demand sides [3]. Group 2: Trusted Data Space - The establishment of a trusted data space is a critical focus area, serving as a virtual computing environment where data providers and users can access and perform computations on encrypted data [4]. - This environment ensures that data is "available but not visible," providing reliable support for the market-oriented circulation of data elements [4]. - The company has already launched several solutions related to the trusted data space, with corresponding practical applications underway [4].

Core Insights - The Beijing government is exploring data security governance for personal, enterprise, and public data to enhance service delivery and convenience [3][4] Group 1: Personal Data - The core issue with personal data is the lack of unified anonymization standards, which creates concerns for data circulation among enterprises [3] - Beijing is collaborating with hospitals to create public datasets from chest X-ray imaging data, aiming to balance patient safety and research needs [3] Group 2: Public Data - Balancing public interest and commercial utilization of public data is a key challenge, with no unified regulations on authorization and pricing [3][4] - The Beijing government has adopted a decentralized authorization approach for public data, particularly in the financial sector, and is considering whether to expand this or shift to centralized authorization [3] Group 3: Enterprise Data - Trust is crucial for enterprise data circulation, and external factors significantly influence its value [5] - Beijing is utilizing blockchain technology to establish a value-added collaboration network to enhance trust among enterprises, particularly in the steel industry where strong interconnections exist [5]

Group 1: Core Insights - The article highlights the significant advancements in artificial intelligence (AI) and data infrastructure in Beijing, with 29 state-owned enterprises actively pursuing 38 key tasks to embrace new opportunities in AI development [1][2][3] - The establishment of a city-level intelligent computing center aims to address the bottleneck in computing power demand, with the Beijing Digital Economy Computing Center planning to provide over 1000P of intelligent computing power through 3600 intelligent computing cabinets [2][3] - The focus on domestic chip commercialization is evident, with the introduction of a platform that integrates over 10 types of domestic chips to validate their performance in various AI applications, thereby reducing verification costs [3] Group 2: Data Flow and Utilization - A new data circulation platform, akin to an e-commerce model, is being developed to facilitate the efficient flow of high-quality data, addressing the supply-demand mismatch in data resources [4][5] - The platform aims to connect at least 10,000 market entities and 30,000 data products by the end of next year, enhancing the social supply of data products and services [5] - State-owned enterprises are also participating in the construction of data asset value realization, with 20 enterprises completing data asset registration and 10 enterprises piloting the entire chain of data asset registration, trading, and application [5] Group 3: AI in Financial Services - The implementation of AI in financial services has led to a 40% increase in efficiency for credit approval processes, significantly improving the quality and speed of service [6][7] - The AI Banking platform developed by Beijing Bank has created over 20 intelligent products and tools, enhancing the overall customer service experience [6][7] - AI-driven customer service has effectively managed a large volume of inquiries, achieving a 92% success rate in responses, thus alleviating the workload on human staff [6][7]