Unimaginable as it may seem, as the dawn of 2025 broke, Nasdaq, nestled in the Western Pacific, was plunged into chaos overnight due to a Chinese AI startup! On January 27, the Nasdaq Composite Index tumbled by 3.07%, and AI giant NVIDIA witnessed its market value erode by a staggering $4.3 trillion in a single day!

Concurrently, NVIDIA founder Jen-Hsun Huang's personal wealth dwindled by over $20.8 billion in the blink of an eye. Oracle's Larry Ellison lost $27.6 billion, Michael Dell $12.5 billion, and Google co-founders Larry Page and Sergey Brin each saw their fortunes shrink by over $6 billion.

Subsequently, President Trump even emphasized at a Republican Party meeting that the ascendancy of the Chinese AI startup DeepSeek should serve as a "wake-up call" for American corporations.

So, who exactly is DeepSeek, and how could it incite such panic in the American capital market?

The True Identity of DeepSeek

DeepSeek's full name is "Hangzhou DeepQuest AI Basic Technology Research Co., Ltd." Its app once topped the free app download charts in both China and the US on Apple's App Store, even surpassing ChatGPT in the US rankings.

What truly astonished everyone was that this AI company was founded by the quantitative asset management giant Quantum Black. In China's capital market, quantitative trading is notorious for being one of the primary culprits behind the market's long-term sluggishness, much to the chagrin of retail investors. This is primarily because quantitative trading leverages AI to incessantly exploit retail investors.

Admittedly, technological development necessitates capital propulsion. Driven by profitability, these quantitative institutions began pouring significant investments into AI research and development. With AI's assistance, they identify arbitrage opportunities by analyzing market sentiment, capital flows, and other data, specifically targeting the irrational fluctuations of retail investors. Quantum Black, with the aid of DeepSeek, has elevated "quantitative investment + AI" to new heights.

Motivated by the profit-seeking nature of capital, DeepSeek has become proficient in AI technology, allowing it to secure more funds to attract top technical talents. The founder of DeepSeek, Liang Wenfeng, is also the founder of Quantum Black, one of the "Four Heavenly Kings" of quantitative private equity.

In 2015, Liang Wenfeng and his Zhejiang University alumni co-founded Quantum Black, aiming to establish China's leading quantitative hedge fund. They deeply integrated mathematics and AI, embarking on an innovative journey in quantitative investment. In 2016, Quantum Black launched its first AI-driven live trading, subsequently fully realizing AI-driven trading strategies. This innovative measure enabled the fund's return rate to far exceed the Shanghai and Shenzhen 300 Index over the same period, driving rapid growth in the company's management scale, which surpassed the 100 billion yuan mark in 2021, ranking among the domestic "Four Heavenly Kings" of quantitative private equity.

As the demand for computing power in AI trading strategies soared, Liang Wenfeng further increased his continuous investment in computing power by leveraging the excess returns from quantitative institutions. These investments have also enhanced Quantum Black's returns in the capital market, leading to the emergence of today's DeepSeek.

Nasdaq in Disarray: DeepSeek's Secret Weapon

The tumult in the American capital market serves as a testament to DeepSeek's prowess. Although the DeepSeek team comprises only 139 members, significantly fewer than OpenAI's 1,200, they have achieved astonishing breakthroughs in large model research and development through their innovative spirit and solid technology.

First and foremost, this is inseparable from the exceptional performance of DeepSeek's products. DeepSeekV2, released in May 2024, achieved cost-effective training and inference with its innovative MLA (Multi-head Latent Attention) and DeepSeekMoE (Mixture of Experts) architectures. DeepSeek-V3, launched in December 2024, stunned the world by outperforming top open-source models in multiple benchmark tests, rivaling top closed-source models like GPT-4, with less than one-twentieth of the development cost.

On January 20, 2025, DeepSeek officially unveiled the DeepSeek-R1 model. This model is on par with OpenAI GPT-4 in tasks such as mathematics, coding, and natural language reasoning, and even surpasses OpenAI in some aspects. DeepSeek stated that R1 extensively employs reinforcement learning techniques during the post-training phase, significantly enhancing the model's reasoning ability with minimal labeled data.

Another standout feature is the extraordinarily high cost-effectiveness of the DeepSeek model. Take the DeepSeek-R1 model as an example; its training cost is only $5.6 million, a fraction of the hundreds of millions or even billions of dollars invested by American tech giants in AI technology. Additionally, the model's query cost is exceptionally low at $0.14 per million tokens, compared to OpenAI's $7.50. In comparison, DeepSeek is 98% cheaper than OpenAI.

The third remarkable feature is DeepSeek's commitment to open-source. DeepSeek not only made the R1 training technology fully public but also distilled six small models and opened them to the community, allowing users to train other models with them. In contrast, OpenAI currently only allows partial open-source and ceased providing API services to China and other countries and regions starting from July 9, 2022.

The China-US AI Rivalry

In recent years, the competition between China and the United States in the field of AI has intensified. The United States has long dominated the AI landscape due to its robust research capabilities, capital investments, and talent pool. China, however, has swiftly emerged as a formidable AI powerhouse, fueled by strong government support, active corporate investments, and the construction of an innovative ecosystem.

The emergence of DeepSeek has unveiled the true potential of Chinese AI! This achievement not only shatters the perception that Chinese AI enterprises rely on imitation but also establishes China's prominence in the global AI competition with its pioneering spirit. The rivalry between China and the United States in AI will undoubtedly become increasingly fierce.

First, there is competition in technology research and development. The United States has taken several measures to safeguard national security and consolidate its advantage in AI technology, including increasing investments in key basic technology areas and tightening controls on the export of advanced semiconductors and AI technologies. China's investments in AI are also on the rise, aiming to secure a leading position in the global competition.

Currently, the United States leads in key hardware areas such as GPUs and AI-specific chips. Simultaneously, leading American AI enterprises have deep roots in the basic and core technology layers, successfully establishing technology clusters in multiple key areas like processor architectures, core algorithms, and system software. Chinese AI enterprises, on the other hand, have amassed abundant patent portfolios in application-layer technologies such as image recognition, speech recognition, and autonomous driving, reflecting China's strengths in technology implementation and commercialization.

Second, there is competition for talent. Currently, 75% of the world's top AI talents hail from the United States and China. American companies like Google, Microsoft, and Meta have attracted a plethora of top AI scholars and talents with their abundant funds and resources. At the same time, the United States has implemented several measures, including frequent interventions in the semiconductor and chip fields, to compete for a significant number of high-end AI research talents from China through various means. For instance, Chinese scientists constitute over a quarter of the OpenAI ChatGPT development team.

China is also not lagging behind, making significant efforts to actively cultivate and attract AI talents. China's prestigious universities have become prominent in the AI field, nurturing a vast pool of outstanding talents. Additionally, China encourages enterprises and research institutions to intensify the introduction and cultivation of AI talents through policy support and capital investments.

Third, there is competition in the industry. The United States boasts numerous top tech enterprises and research institutions, such as Google, Microsoft, and Amazon, which have heavily invested in AI algorithms and chip research and development. These enterprises have achieved remarkable milestones in the research and application of AI technology, propelling the rapid growth of the American AI industry.

Simultaneously, China is also actively strengthening the layout and development of the AI industry. Although China is a relative newcomer in the AI field, the government and major tech companies have invested heavily in building infrastructure such as supercomputers, cloud computing centers, and data centers. China has developed some of the world's fastest supercomputers, like the "Tianhe" and "Sunway" series, and the cloud computing market is also expanding rapidly. Furthermore, China has fostered the rapid development of the AI industry by issuing a series of policies to encourage enterprises and research institutions to innovate in the AI domain.

Fourth, there is competition in the market. AI enterprises in both China and the United States are vigorously competing for global market share. With its innovative algorithms and efficient model design, China's DeepSeek has outperformed OpenAI's ChatGPT in numerous tests and boasts a lower design cost, significantly impacting the American tech industry and inevitably capturing a larger market share.

The United States is also supporting its AI development by initiating AI infrastructure investment projects like "Stargate" to compete with China and seize the AI market.