The ripple effects of DeepSeek continue to unfold.

Investor Network - Han Yijia

Amidst the festive reunion moments of the Spring Festival, a domestically developed AI platform captured the eager attention of both Chinese and international audiences.

Just prior to the festival, DeepSeek (Depth Exploration) burst onto the scene, shaking up the AI industry with its stellar performance and cost-effective development.

On January 27, US time, US stocks plunged at the opening bell, with the technology sector bearing the brunt of the losses. US chip giants NVIDIA and Broadcom saw their share prices plummet by approximately 17%, AMD's fell by 6%, and Microsoft's by 2%. Furthermore, derivatives in the AI field, such as power suppliers, also took a heavy hit, with American Electric Power's share price down 21% and Vistra's down 29%.

DeepSeek's emergence has not only impacted the US stock market but also stirred up the domestic market. Related company Daily Interactive hit its daily trading limit on January 27, the last trading day before the Spring Festival. Prior to this 20% surge, the company's share price had also risen by nearly 20% within a week.

Bolstered by the positive impact of DeepSeek, the Hong Kong stock AIGC concept index has continued to climb. According to Tonghuashun iFinD data, the index has cumulatively risen by 27.23% over 8 trading days since January 21st.

Following the pre-holiday "surge," the impact of DeepSeek remains undiminished. During the Spring Festival, DeepSeek not only trended on search engines with the topic "recruiting with an annual salary of one million yuan" but also gained a substantial number of domestic and foreign users despite facing malicious attacks.

According to research reports from international investment banks, DeepSeek's current valuation has soared to $10 billion. Observing the market response, it is evident that the changes brought about by DeepSeek extend beyond the AI industry alone.

“A Mysterious Eastern Force”

DeepSeek, formally known as Hangzhou Depth Exploration Artificial Intelligence Basic Technology Research Co., Ltd., was founded on July 17, 2023, specializing in the development of advanced large language models (LLMs) and related technologies. This homegrown startup technology company has crafted an open-source model comparable to OpenAI, led by a "refined" startup team.

Public information reveals that DeepSeek's founder, Liang Wenfeng, is a post-85s individual who was admitted to Zhejiang University's undergraduate program in Electronic Information Engineering at the age of 17 with the top score in the college entrance examination and went on to pursue a master's degree in Information and Communication Engineering at Zhejiang University in 2007. In 2013, Liang Wenfeng co-founded Hangzhou Jacobi Investment Management Co., Ltd. with his Zhejiang University classmate Xu Jin, and two years later established Hangzhou QuantBox Technology Co., Ltd., dedicated to quantitative investment leveraging mathematics and artificial intelligence.

In 2021, QuantBox's asset management scale surpassed the 100 billion yuan mark. In 2023, Liang Wenfeng announced his official foray into the field of general artificial intelligence and founded Depth Exploration DeepSeek, with a focus on creating true human-level artificial intelligence.

At the end of 2024, DeepSeek unveiled its new generation large language model V3, simultaneously announcing its open-source availability. Test results indicate that it surpasses several mainstream open-source models in various evaluation metrics while also offering a cost advantage.

On January 20, 2025, on the opening day of the World Economic Forum's 2025 Annual Meeting, DeepSeek released its latest open-source model R1, once again capturing global attention in the AI realm.

According to the company, the R1 model has achieved significant technical breakthroughs—employing pure deep learning methods to enable AI to spontaneously develop reasoning abilities, performing on par with the official version of the o1 model from the US OpenAI Research Center (OpenAI) in tasks such as mathematics, coding, and natural language reasoning. This model also continues the company's advantage of high cost-effectiveness.

It is understood that the training cost of Depth Exploration Company's R1 model is a mere $5.6 million, a fraction of the hundreds of millions or even billions of dollars invested by US technology giants like the OpenAI Research Center, Google, and Meta in AI technology.

In Silicon Valley, DeepSeek has long been hailed as a "mysterious force from the East." The release of the R1 model has thoroughly shaken the entire AI industry, not only impacting the prices of US technology stocks but also topping the global download charts on Apple's App Store and Google Play Store for consecutive days.

DeepSeek-R1 is recognized as one of the most advanced large language models currently available, offering high-quality language processing capabilities. In essence, DeepSeek has employed a "smarter" algorithm, reducing AI training costs by nearly 60% while achieving or even surpassing the performance of similar models. This "cost-effectiveness" directly addresses the pain points of the industry.

Liu Wei, director of the Human-Computer Interaction and Cognitive Engineering Laboratory at the School of Artificial Intelligence, Beijing University of Posts and Telecommunications, explains that DeepSeek's biggest advantage lies in the improvement and optimization of algorithms, saving computing power and data volume. If OpenAI was previously known for "miracles through brute force," now DeepSeek demonstrates that "miracles can also be achieved with less force"—smaller computing power can achieve remarkable results with innovative methods.

Analysts believe that this progress has shattered the long-standing path dependency of "stacking computing power" in the global AI industry. The DeepSeek model is expected to spark a wave of innovation, driving continuous progress in global AI.

Huatai Securities stated that DeepSeek's advantages of low cost and high performance have sparked discussions in the industry about capital expenditures and application scenarios, which may prompt investors to reassess the technological potential of Chinese technology enterprises, thereby catalyzing a revaluation of Chinese and American technology stocks.

Zheshang Securities also emphasized that the disruptive nature of DeepSeek-R1 lies in its successful catch-up with overseas models, particularly the OpenAI model representing the world's leading AI level, for the first time by a Chinese model. By optimizing algorithms to achieve AI equity, it breaks the established paradigm of "miracles through brute force" in computing power and chips, which will prompt a revaluation of global AI assets. Zheshang Securities believes that the DeepSeek model is fueling a global AI wave, driving continuous AI progress.

Recently, multiple platforms, including Baidu Intelligent Cloud, Huawei Cloud, Alibaba Cloud, Tencent Cloud, 360 Digital Security, and ZStack Cloud, have announced the launch of the DeepSeek large model. Overseas, global technology vendors such as Amazon AWS, Microsoft Azure, and NVIDIA have successively announced their integration with the DeepSeek model.

For the AI industry, the impact of DeepSeek is undeniably profound and widespread.

Navigating Commercialization Paths

DeepSeek's impact on the entire industry chain is also ongoing.

On January 27, DeepSeek ignited a wave of enthusiasm in the A-share market. Multiple concept stocks related to DeepSeek recorded gains on this trading day. According to Wind data, on January 27, the Wind DeepSeek Partner Index surged by over 13% in the morning session and closed up 10.96%.

Among them, Daily Interactive and Zhuochuang Information both closed with a 20CM daily limit, while stocks such as Tuersi, Bingxing Technology, and Feilixin witnessed significant gains. Zhejiang East, which participated in DeepSeek's angel round of funding, saw its share price rise by 38.2% within the last 10 days; Huajin Capital, which participated in DeepSeek's Pre-A round of funding, saw its share price rise by 21.7% within 10 days.

Many companies have capitalized on this "wave of wealth." Daily Interactive announced on February 4 that the company had accessed the relevant version of DeepSeek earlier and completed various assessments and privatized deployment using its own computing power. Daily Interactive stated that as the industry enters the era of large models, the company will embrace the open technical ecosystem of the DeepSeek large model.

With DeepSeek being a "hot topic," many companies have leveraged it, while others have rushed to clarify their positions. Zhuochuang Information came forward to refute rumors, stating that the company has not yet established business cooperation with DeepSeek.

Despite denials from listed companies, investors' enthusiasm for DeepSeek and its related industry chain remains undiminished. Information about enterprises related to DeepSeek's business has been constantly updated, and many investors have inquired on platforms whether other listed companies are utilizing the DeepSeek large model.

It is undeniable that DeepSeek's success has "ignited hope" in many enterprises. Investor Zhejiang East has forecast that its net profit in 2024 will reach 947 million yuan, with an expected year-on-year increase of over 75%. Huajin Capital has also shown a strong growth trend, expecting to achieve a profit of 145 million to 185 million yuan, with a year-on-year increase of over 95%.

However, among these "DeepSeek concept stocks," many companies have issued loss forecasts. Daily Interactive has predicted a net loss attributable to shareholders of the parent company for 2024 ranging from 61.5 million yuan to 91 million yuan, with a net profit after deducting non-recurring gains and losses expected to be between 16 million yuan and 23.5 million yuan.

The future commercial development of DeepSeek is also worth monitoring. Given its adoption of "open source," whether it can explore advanced services to expand its business scope in the future remains a key focus of the market. In the long run, how "small-cost"-driven DeepSeek can leverage key technologies and large-scale computing power is a question that needs to be contemplated.

Yet, there is no doubt that the emergence of DeepSeek has broken the monopoly of traditional AI giants and reshaped the competitive landscape of the global AI market. (Produced by Sifinance) ■

Source: Investor Network