The AI sector grew by leaps and bounds in recent years, fueled by advancements in AI tech, the demand for popular chatbots like ChatGPT and Gemini, and the millions that have been invested in the sector by investors across the globe. Now, the exact opposite has happened – China-origin DeepSeek’s foray into AI has resulted in a massive upheaval in the global tech sector – the company’s low-cost, high-performance AI models have now challenged established players in the sector and triggered an unprecedented sell-off in tech stocks, wiping out billions of dollars in market value.
DeepSeek, the two-year-old Hangzhou-based startup, recently released its DeepSeek-V3 and DeepSeek-R1 models. These AI platforms, according to the company, rival the capabilities of leading Western competitors like OpenAI’s ChatGPT but operate at a fraction of the cost. What makes these developments all the more enticing is that DeepSeek’s AI models were reportedly developed on Nvidia’s less powerful H800 chips at a cost of under $6 million—significantly less than what competitors typically spend. In other words, you can have fast, efficient performance at a fraction of the cost.
The DeepSeek-R1 model, introduced just a week ago, is said to be 20 to 50 times more cost-efficient than similar offerings from industry giants. Independent third-party evaluations have corroborated the company’s claims, confirming that DeepSeek’s models outperform in several key areas. Moreover, the decision to make these models open-source means that developers worldwide have access to, and can deploy, these tools without substantial investment on their part.
DeepSeek’s gain has proved to be the detriment of other tech companies, though. Nvidia, a dominant force in AI hardware, saw its stock plunge by nearly 17%, translating to a loss of $593 billion in market capitalization. This marked the largest single-day loss in the history of Wall Street, surpassing even its previous record of $279 billion in September. “DeepSeek demonstrates an alternative path to efficient model training than the current arm’s race among hyperscalers by significantly increasing the data quality and improving the model architecture. DeepSeek is now the lowest cost of LLM manufacturing, allowing frontier AI performance at a fraction of the cost with 9-13x lower price on output tokens vs. GPT-4o and Claude 3.5,” Morgan Stanley commented on the development.
Other semiconductor companies were similarly affected. Broadcom’s shares fell by 17.4%, while Marvell Technology suffered a 19.1% decline. The Philadelphia Semiconductor Index, a key benchmark for the sector, dropped by 9.2%, its steepest decline since the onset of the pandemic back in 2020. According to analysts, this widespread fall can be attributed to fears that the demand for high-performance AI chips could be significantly reduced by the rise of more efficient alternatives like DeepSeek’s models.
The Nasdaq Composite Index, heavily populated with tech and AI-focused firms, fell by 3.1%, marking one of its most significant single-day declines in recent years. At its lowest point during the trading session, the index had lost more than $1 trillion in valuation. Similarly, major Big Tech companies like Microsoft and Alphabet, which have invested heavily in proprietary AI technologies, also experienced notable declines. Microsoft, a key backer of OpenAI, saw its shares drop by 2.1%, while Alphabet, Google’s parent company, fell by 4.2%. Outside the U.S., global markets mirrored this downturn. Japan’s SoftBank Group closed 8.3% lower, and European semiconductor firms like ASML reported losses of over 7%.
