Hong Kong-listed DeepSeek concept stocks saw a strong performance today, with most stocks surging. As of press time, Baidu Group-SW was up 9.18%, Kingsoft Cloud was up 5.03%, and Alibaba-W was up 4.58%. This rally is driven by several factors: Baidu's Wenxin large language model (LLM) technology upgrades, the surge in computing power demand fueled by Tencent's accelerated deployment of LLMs, and a positive confluence of policy support and market sentiment, leading to sustained heating up of the technology sector.
Baidu's LLM Technological Breakthrough
On March 16, Baidu officially launched Wenxin LLM 4.5 and Wenxin LLM X1. Both models are now freely available to users on the Wenxin Yiyan website, with plans to open-source the 4.5 series on June 30.
According to official statements, Wenxin 4.5 shows significant improvements in programming capabilities and multilingual reasoning, while X1 focuses on vertical scenario optimization. Wenxin 4.5 has been integrated into Baidu Smart Cloud's Qianfan large model platform, allowing enterprise users to directly access its API. X1 will also be integrated soon, further expanding the scope of technology applications.
This launch is seen as a key move for Baidu to solidify its position in the AI field. Market analysis suggests that the free and open strategy will accelerate the penetration of the large model ecosystem, while the open-source plan may promote the development of the developer ecosystem, bringing long-term benefits to Baidu's cloud business and advertising monetization capabilities. Boosted by this news, Baidu's US stock surged 9% overnight, and its Hong Kong stock continued its strong performance today.
Tencent's Increased Investment in the LLM Race
According to market reports, Tencent recently placed an order with Nvidia for H20 chips worth billions of yuan, primarily to support the integration of DeepSeek's large language model into WeChat. Due to the concentrated order, H20 chips are experiencing a temporary shortage in the Chinese market. Although Tencent has not directly responded, industry insiders believe this reflects its determination to accelerate the implementation of large language models.
Previously, the DeepSeek concept has repeatedly ignited the Hong Kong stock market. In early March, due to DeepSeek's plan to release its next-generation AI model R2 ahead of schedule, related stocks such as GDS Holdings and XD Inc. saw daily gains exceeding 15%. Tencent's recent purchase further strengthens market expectations for the computing power industry chain, coupled with the Nvidia GTC conference, boosting the popularity of AI chips and cloud computing.
Policy and Capital Synergy: Tech Stocks Become the Market's Main Trend
Recently, the policy side has continuously released positive signals. The government work report clearly stated the "AI+" action plan, supporting the widespread application of large models; the State Council Development Research Center also specifically mentioned DeepSeek and other companies, acknowledging their breakthroughs in the AI field. At the same time, southbound funds are accelerating their inflow into the Hong Kong tech sector. On March 17, Alibaba's US stock rose by more than 5%, and the Chinese ADR index hit a three-year high, indicating a warming-up of foreign investment demand for Chinese technology assets.
From a market structure perspective, DeepSeek concept stocks cover multiple areas including cloud computing, semiconductors, and smart terminals, largely overlapping with the constituent stocks of the Hang Seng Tech Index. Today, the Hang Seng Tech Index broke through the 6000-point mark, and the widespread rise of leading technology stocks highlights the capital's preference for growth sectors.
Institutions Warn of Short-Term Volatility
Despite the high popularity of AI-related stocks, institutions point out that the commercialization progress of large models may fall short of expectations; if companies like Baidu and Tencent fail to continuously launch innovative applications, market sentiment may be dampened; furthermore, fluctuations in the computing power supply chain, and potential changes in Nvidia's chip supply due to international circumstances, will directly affect the progress of domestic large model R&D.
