On February 10, 2025, a powerful force seemed to have ignited the A-share market as DeepSeek-related stocks surged once again. Amidst the clamor and excitement, the market sentiment reached a peak, driven by the groundbreaking announcement that the three major telecommunications companies had fully integrated with the DeepSeek open-source large model. This integration has transformed the landscape of artificial intelligence (AI) investments, and the ripple effects are immediate and significant for investors and companies alike.

The share prices of the three telecommunications giants soared as if injected with a potent booster. China Telecom’s stocks skyrocketed, reaching their limit and setting a historical record, underscoring the market's optimistic outlook on this collaboration; China Unicom did not lag behind, also surging more than 7%, reflecting significant market vitality; China Mobile garnered attention, with gains exceeding 4%. Besides these major players, many associated stocks also rose sharply. Megvii Technology hit a remarkable six consecutive trading limits, becoming the focal point of the market's interest; Qingyun Technology and Hangzhou Steel Group each recorded four consecutive limits, showcasing impressive upward momentum; YK Technology and Wan Shun New Materials saw their stock prices soar by over 10%, further heating up the market. By the time this report was drafted, while the gains had relaxed somewhat, the fervor of this movement remained striking.

Following the Spring Festival, the first trading week saw DeepSeek stocks maintaining their lead in the A-share market. From February 5 to 10, specific segments such as AI data, computing power leasing, and edge-side AI rose sharply. Financial inflows reflected this trend; an AI ETF saw a net inflow of nearly 20 million yuan over the past week, indicating strong investor interest and demand in the AI sector, further driving the market's momentum.

The surge in market activity correlates closely with the commercialization advancements of the DeepSeek model. On February 6, China Mobile, China Unicom, and China Telecom jointly announced the successful integration of DeepSeek-R1 and V3 models on their cloud computing platforms. This strategic initiative is pivotal, as it enhances the practical applications of the DeepSeek model, leveraging the telecommunications companies' formidable cloud computing capabilities. Looking ahead, the partners aim to collaborate on deploying this model across various sectors such as smart cities and industrial quality inspections, injecting fresh energy into their developmental trajectories. Previously, companies like Baidu Cloud, Tencent Cloud, and Huawei Cloud also introduced cost-effective solutions based on the DeepSeek model, employing time-limited free strategies, which act as a catalyst for engaging market needs and encouraging wider adoption of DeepSeek's offerings.

The explosive growth of DeepSeek is not merely fortuitous; it is a result of significant technological innovations. Reports indicate that the DeepSeek-V3 model has achieved remarkable advancements through intricate algorithmic optimizations, cutting training costs down to 10% of the industry average, enabling more businesses to participate in AI developments affordably. Moreover, its inference efficiency has improved by more than three times, achieving performances in critical tasks such as code generation and multimodal understanding that rival GPT-4. This technological leap is poised to reshape the competitive landscape in the global AI arena.

Experts analyze that the “low-cost + high-performance” combination of DeepSeek is groundbreaking. This development allows small and medium-sized enterprises to bypass longstanding constraints regarding computing power, enabling them to engage actively in the AI application development wave. Forecasts suggest that by 2025, the pace of commercialization for China's AI large models could accelerate by 40%, significantly bolstering the country's AI industry. Furthermore, DeepSeek's influence is extending into international markets; global giants like Amazon and NVIDIA have announced their adoption of the DeepSeek model, indicating its widespread recognition and application on a global scale. Meanwhile, reports indicate that Meta is urgently initiating “anti-DeepSeek” technological initiatives internally, reflecting the model's impact and the competitive pressure it exerts on industry behemoths.

However, as enthusiasm in the market continues to rise, subtle risk signals have begun to surface. In the capital market, several companies have issued statements clarifying that they do not hold equity in DeepSeek's parent company nor have they provided training data, emphasizing that business cooperation has many uncertainties. This situation suggests that some stocks tied to DeepSeek may be experiencing excessive speculation, leading to price increases without tangible business backing. Moreover, the rise in phishing websites impersonating DeepSeek has surged, with scammers advertising “income of one million a month” to sell related courses and exploit investors. These fraudulent activities not only harm investor interests but also disrupt market order.

Industry experts have also pointed out deeper challenges facing DeepSeek. Regarding core computing power, China still faces difficulties due to restrictions on high-end chips. Unless viable Chinese substitutes like Huawei Ascend and Cambricon achieve substantial breakthroughs, the long-term competitiveness of DeepSeek may be seriously threatened. Additionally, a relentless talent competition is emerging. According to recruitment data, the average monthly salary of AI algorithm engineers has soared to 80,000 yuan, and reports indicate that DeepSeek interns are earning over 1,000 yuan a day. This fierce competition for talent not only escalates operational costs for companies but also poses challenges for the healthy development of the sector.

The capital frenzy surrounding DeepSeek represents a vivid illustration of the democratization of AI technology. As large models transition from previously unattainable “technological singularities” to broadly utilized “infrastructures,” innovation at the application layer will become critical in determining the outcome of this rapidly evolving landscape. Nevertheless, amidst this exuberance, finding a balance between valuation bubbles and genuine value, and avoiding blind investment following trends, is a critical reflection and challenge that every investor must confront amidst this ongoing transformation.