Nvidia, a leading US chipmaker, experienced a sharp drop in its stock price after a Chinese AI breakthrough stirred competition concerns. The development, led by Hangzhou-based startup DeepSeek, introduces a cost-efficient generative AI model that could disrupt the global AI landscape.
On January 20, DeepSeek revealed its R1 model, capable of competing with top-tier U.S. AI systems. Notably, the startup trained the model for just $5.6 million, a fraction of the estimated $100 million cost for OpenAI’s GPT-4.
Market observers noted that DeepSeek’s R1 model has reshaped expectations in the AI industry. By delivering high-performance results on lower-cost hardware, the model challenges the dominance of US companies like OpenAI. David Sacks, the White House Crypto and AI advisor, wrote:
“DeepSeek R1 shows that the AI race will be very competitive and that President Trump was right to rescind the Biden EO, which hamstrung American AI companies.”
Scale of Nvidia’s Loss
By January 27, the news triggered a steep sell-off in Nvidia’s shares, which fell over 15% during the day’s trading. This drop wiped out approximately $600 billion in market value, marking one of the most significant losses in tech market history.
Crypto and financial analyst Noelle Acheson highlighted the staggering scale of Nvidia’s market cap reduction, noting that it exceeded Ireland’s annual GDP. She said:
“NVIDIA accounts for four of the top-five market cap losses ever and the scale of today’s is quite something—in 24 hours it lost more than the annual GDP of Ireland.”
Market analysis platform Kobeissi Letter noted that Nvidia has erased all gains since October 2024, losing over 20% in just two trading sessions. It added:
“This was the largest public company in the world last week, erasing one fifth of its value in 2 days.”
Meanwhile, the broader AI sector also saw ripple effects of Nvidia’s losses. AI-related cryptocurrencies, such as AIOZ Network, Render, and Arweave, suffered declines of 10.2%, 9.3%, and 9.4%, respectively. However, Bittensor bucked the trend, recording a 5.6% gain during the reporting period.
Nvidia Unperturbed
Despite the financial blow, Nvidia praised DeepSeek’s R1 model. A company spokesperson called it an excellent example of “Test Time Scaling,” a method that builds on existing technologies while complying with export regulations.
The spokesperson said:
“DeepSeek’s work shows how new models can be created, leveraging widely-available models and compute that is fully export control compliant.”
Nvidia also emphasized the ongoing role of its hardware in advancing AI. The spokesperson pointed out that “inference still depends heavily on Nvidia GPUs and advanced networking solutions,” which suggests that the demand for their products in deploying models like R1 remains high.
