The A-shares market in China is currently experiencing a significant surge centered around the DeepSeek concept stocks. Despite the evident enthusiasm in the market, analysts are cautioning that cautionary risk signals are starting to emerge. This recent wave of excitement reached a peak on February 10, 2025, driven primarily by the announcement from the country's three major telecommunications companies, which revealed that they would fully integrate the DeepSeek open-source model.

The stock prices of these telecom giants shot upward, with China Telecom hitting its daily limit and achieving a historical high. This impressive price surge signifies the market's optimistic outlook regarding the potential expansion of their services following the integration of the DeepSeek model. Concurrently, China Unicom saw its stock rise over 7%, while China Mobile reached an increase exceeding 4%. Although the gains slightly moderated as the trading continued, the fluctuations in stock prices of these three titans captured the attention of many investors. Beyond the telecom sector, companies such as Meige Intelligent demonstrated remarkable growth, achieving six consecutive gains. Others like Qingyun Technology and Hangang Shares also performed strongly with four consecutive increases. Stocks like Youkede and Wanshun New Material enjoyed peaks exceeding 10%, showcasing how the entire DeepSeek sector benefitted from this momentum.

The week commencing with the Chinese New Year saw the DeepSeek concept stocks continue to lead the market. Between February 5 and 10, sub-sectors like AI datasets, computing power leasing, and edge AI experienced substantial gains. Notably, the AI-focused exchange-traded fund (ETF) had nearly 20 million yuan in net inflows over the past week, indicating robust investor interest in this segment.

This latest market uptrend correlates closely with the commercial advancement of the DeepSeek open-source model. On February 6, China Mobile, China Unicom, and China Telecom collectively announced that they had successfully integrated the DeepSeek-R1 and V3 models into their cloud computing platforms. This move is expected to enable these technologies to scale and be deployed across various scenarios, including smart city developments and industrial quality inspections. Prior to this joint announcement, several tech giants such as Baidu's intelligent cloud, Tencent's cloud services, and Huawei's cloud offerings unveiled competitively priced solutions based on the DeepSeek model, with limited-time free strategies further stimulating market demand.

This technological breakthrough is poised to shift the landscape of global AI competition. Analysts speculate that the “low-cost + high-performance” combination of DeepSeek empowers small and medium-sized enterprises to bypass the computational power bottlenecks that have limited their ability to engage directly in AI application development. Consequently, it is anticipated that the commercialization of AI large models in China will accelerate by up to 40% by 2025. The international market has also felt the tremors from this trend, as companies such as Amazon and NVIDIA have signaled their adoption of the DeepSeek model. Reports suggest that within Meta, there is a rush to initiate “anti-DeepSeek” technological efforts.

Despite the rising market fervor, cautionary indicators are becoming evident. In the capital markets, multiple companies have released announcements clarifying their lack of equity ownership in DeepSeek’s parent company, as well as denying any roles in providing training datasets, highlighting existing uncertainties in business collaborations. Additionally, recent monitoring by Qihoo 360 has revealed a dramatic increase in phishing websites that impersonate DeepSeek, with some course vendors promising massive monthly incomes as bait to attract potential investors.

Industry experts have highlighted that essential computational power remains constrained due to high-end chip bans, posing a long-term competitive challenge for DeepSeek if domestic alternatives such as Huawei’s Ascend or Cambricon fail to achieve breakthroughs. Furthermore, a talent war has quietly escalated as well: data from recruitment platforms reveal that the average salary for AI algorithm engineers has surged to 80,000 yuan monthly, with the DeepSeek development team allegedly offering over a thousand yuan in daily wages for interns.

The capital frenzy triggered by DeepSeek fundamentally mirrors the democratization of AI technology. As large models transition from being technological singularities into essential infrastructures, innovation at the application layer will emerge as the deciding factor for success. Nevertheless, investors must grapple with the challenge of finding a balance between valuation bubbles and genuine value within this rapidly evolving landscape.