Entering China
On March 3, local time, as a leading technology company in the United States, Nvidia (NVDA) fell 8.69% to $114.06 on the same day. The highest intraday drop once exceeded 10%. The company’s market capitalization has shrunk to $2.79 trillion, and Monday’s drop wiped out another $265 billion from Nvidia’s market capitalization.
It has fallen nearly a trillion dollars from its peak market capitalization
In the longer term, Nvidia’s current market capitalization has shrunk by about $900 billion from its peak of $3.66 trillion on January 6. However, Nvidia executives said the company would benefit from the development of “reasoning AI” because it would increase the demand for reasoning computing.
At the same time, other AI-related stocks also fell sharply. In addition to Nvidia, Broadcom fell more than 6%, Amazon fell more than 3%, and Tesla, Google, Microsoft, etc. also fell more than 2%. U.S. chip stocks also fell broadly, with the Philadelphia Semiconductor Index down 4.01%, ARM down more than 8%, Marvell Technology down more than 6%, and Intel down more than 4% 。
JPMorgan Chase reported that NVIDIA’s two major customers, Amazon and Marvell, have cut CoWoS orders, reflecting the slowdown in downstream AI computing demand. At the same time, its core business areas are facing increased competition: AMD has launched a new generation of GPU products, while giants such as Google and Meta have accelerated the process of developing their own chips.
In addition, Nomura data showed that Nvidia’s put option positions in the $115-$130 exercise price surged, resulting in the accumulation of negative gamma values among traders. This position structure means that once the stock price falls below a key support level, option sellers may be forced to accelerate the sell-off to hedge risk, further amplifying price volatility.
Jordan Klein, a trading desk analyst at Mizuho, said in a media interview that it was “a continuation of last week’s market move” and described the decline as a “cooling down” for AI trading.
The performance is bright, but the stock price fluctuates downward
In fact, Nvidia’s stock price has been in a volatile downward phase recently. After the market on Wednesday Eastern time, Nvidia announced its fiscal year 2025 (corresponding to the fourth quarter) and full-fiscal year performance data for the fourth quarter (corresponding to the fourth quarter) and the full fiscal year, as well as the performance guidance for the first fiscal quarter of fiscal year 2026.
In the fourth quarter of fiscal year 2025, Nvidia achieved revenue of $39.331 billion, a significant increase of 78% year-on-year, far exceeding market expectations of $38.05 billion, demonstrating its strong strength in the market; Net profit reached $22.091 billion, an increase of 80% year-on-year For the full year of fiscal year 2025, Nvidia achieved a major leap forward, entering the “100 billion dollar club” in revenue for the first time, achieving a total revenue of 1304.97 for the year100 million US dollars, a year-on-year increase of 114%; Net profit reached 72.880 billion US dollars, a year-on-year increase of 145%.
However, the reporter noticed that as soon as this financial report was released, it made waves in the market. Nvidia’s stock price did not rise but fell, and in after-hours trading after the earnings report, Nvidia’s stock price fell 8.5%, and its market value evaporated by 274 billion US dollars (about 1.98 trillion yuan) overnight, and then the stock price still fluctuated lower.
More importantly, the impact from Chinese artificial intelligence startup DeepSeek is an important event that cannot be ignored, it has made waves in the AI industry, and also brought unprecedented challenges to Nvidia. The high-performance open-source inference model R1 has been put into the AI market like a blockbuster, causing a strong shock in the market. With its unique advantages, this model has broken the original market pattern and allowed many AI companies to see new possibilities.
In the face of the impact of DeepSeek, Huang has expressed his views on several occasions. He stressed that the impact of DeepSeek on Nvidia stock stemmed from a misunderstanding by investors. Huang believes that DeepSeek’s R1 open-source inference model is “exciting” because it will not end the need for AI computing, but will expand and accelerate the market’s quest for more efficient AI models. In his view, pre-training is only the basic stage of AI, and post-training is the core part of AI, and AI can solve practical problems through reinforcement learning and other methods.
