The article discusses Anker's strategy of deploying computing power locally (edge computing) amid the general trend of AI moving to the cloud. This reflects the demand for low latency and privacy in AI applications, and the rise of edge AI.
DeepSeek is hiring about 150 people, focusing on senior engineers with 2-10 years of experience, covering roles such as backend development and agent elastic computing. This indicates continued investment in model R&D and engineering.
Huawei's Intelligent Automotive Solution BU announced that its Qiankun ADAS and Harmony Cockpit installations, along with Qiankun App users, have simultaneously reached 2 million. The cumulative mileage of Qiankun ADAS has exceeded 42.7 billion kilometers, with 6.4 million potential collisions avoided. Huawei executives stated that they will invest 70-80 billion yuan in computing power R&D over the next five years.
Xiechuang Data (300857.SZ) announced that its computing resources have not yet been integrated into the national computing network's unified scheduling system. This statement may affect market expectations regarding its AI computing power initiatives.
This article reports on the trend of Chinese data centers migrating to western regions rich in power resources to address electricity shortages caused by surging AI computing demand. It mentions Inner Mongolia, Ulanqab, Zhongwei and other locations as hotspots for data center siting, and discusses multi-party efforts under the 'East Data, West Computing' policy.
The article reports that 6G commercialization is imminent, which is expected to bring new benefits to the computing power sector. The development of 6G is anticipated to further drive demand for high-performance computing and AI infrastructure.
A research report on investment opportunities in China's computing network construction has been released, highlighting a trillion-yuan market potential. The report emphasizes computing infrastructure as the cornerstone of the digital economy, driving development across related industrial chains.
The article notes that China's AI computing power industry is shifting from customized to standardized infrastructure construction, with the industry calling for standardized platforms akin to a 'transaction fulfillment foundation' to facilitate computing power trading and deployment. This reflects a maturing trend in the AI infrastructure market, but no specific investment amounts or company financials are mentioned.
Shenhao Technology stated that it has ample orders on hand, but its computing power leasing business has not yet formed orders. This indicates that the company's foray into AI computing power is still in early stages and has not generated actual revenue.
The article reports on the promotion of AI WAN (Intelligent IP Wide Area Network) in China, particularly its deployment progress in Guangdong. Using the metaphor of moving from 'showroom' to 'commercial housing', it indicates that the technology is transitioning from pilot to large-scale commercial use, accelerating the intelligent upgrade of network infrastructure.
A subsidiary of Yakang Co., Ltd. signed a high-performance computing cloud service agreement worth 1.188 billion yuan. The agreement involves providing high-performance computing cloud services, indicating sustained market demand for computing power.
Nanjing Big Data Group has launched a Token store that bills by token, reportedly saving hundreds of thousands of yuan in computing costs per project. The model aims to reduce AI application costs by sharing computing resources.
Hebei Broadcasting and Television reports that Hebei Province is leveraging computing-electricity synergy and other measures to build a smart computing cluster around Beijing. This aims to capitalize on its geographical advantage to absorb spillover computing demand from Beijing and boost local digital economy.
36Kr recommends five early-stage AI projects weekly, covering AI compute scheduling, AI hardware DaaS, AI glasses, supply chain intelligence, and blue-collar recruitment AI. These projects aim to lower AI adoption barriers and improve efficiency, but lack specific financial data.
At its September 7 event, Huawei unveiled several new products, including the Mate XT 2 tri-fold phone powered by the Kirin 9050 Pro, the Pura X View, and more, along with the HarmonyOS 7 system. The system emphasizes immersive experiences, AI assistance, and security features, with a 15% performance boost and 36 minutes more battery life.
The breakout of generative AI is reshaping corporate models and global markets. As information becomes ubiquitous and technological iterations accelerate, established competitive moats face constant redefinition. Yet for Zhong Zhaomin, founder of Oriental Marathon Investment Management, the AI revolution does not render value investing obsolete; it transforms its methods while leaving its core principles intact.
The report reveals that compute usage for office agents is highly concentrated, with the top 20% of users consuming the majority of resources. This may reflect deep reliance on AI tools among power users, but also suggests an imbalance in compute allocation.
The article reviews quality vendors in the AI inference computing track for 2026, highlighting the steady rise of domestic specialized brands. It focuses on the improving competitiveness of Chinese AI inference chip and computing power suppliers, but does not provide specific financial data or investment figures.
The article points out that with the rapid iteration of AI computing power demand, data center cooling technology is shifting from air cooling to liquid cooling, presenting development opportunities for the liquid cooling sector. Investment institutions believe this trend will drive growth in the related industrial chain.
CAICT has initiated a call for test tasks for its 'Fangsheng' agent benchmark, aiming to build a layered testing framework and advance agent testing toward comprehensive capability measurement, process analysis, and problem diagnosis. This move seeks to enhance the benchmark's professionalism, representativeness, and industrial applicability by pooling high-quality task resources.