Latest Search
Quote
| Back Zoom + Zoom - | ||||
|
<Research>G Sachs: DeepSeek's Computing Costs Significantly Lower than Global Peers; Data Centers Expected to Benefit from Public Cloud/ AI Computing Demand
Recommend 39 Positive 68 Negative 26 |
||||
|
||||
|
Over the past week, a number of companies released foundation models, including DeepSeek's R1, ByteDance's Doubao-1.5 Pro and Moonshot's Kimi k1.5 foundation models, Goldman Sachs released a research report saying. As Chinese companies continue to make progress in AI, they attracted the close attention of investors in terms of benchmark compared with global peers. The above companies posted excellent performance, and their training/ computing capability requirements are significantly lower than their global peers. Meanwhile, Goldman Sachs noted that Chinese open-source models, including DeepSeek's R1/V3 and BABA-W (09988.HK)'s Qwen have drawn widespread developer interest since their launch due to their transparency and significantly lower pricing per token compared to global models. On stocks, in terms of applications, the broker continued to see TENCENT (00700.HK) as the most capable of launching To-C AI agent applications, as WeChat is both social and transactional. For cloud/data centers, Goldman Sachs noted geopolitical uncertainties around chip/ computing capacity constraints and progress in training/ inference cost optimizations. However, the broker expected the cloud businesses of internet giants (BABA-W is the largest listed company in China in the related business), cloud computing mega providers and data centers such as GDS Holdings (GDS.US) and VNET Group (VNET.US) to benefit from the continued growth in demand for public cloud and AI computing, which will be fueled by increased AI adoption. AAStocks Financial News |
||||
Copied to Clipboard
 |
||||
