Nvidia states that DeepSeek's success demonstrates competitive AI models can be created under US export restrictions.

Nvidia has acknowledged China's DeepSeek artificial intelligence model amid a significant market downturn. DeepSeek's open-source AI model, developed for under $6 million, reportedly outperformed leading U.S. models like those from OpenAI, triggering a broader tech selloff. The market reaction was substantial: Nvidia shares fell 17% and AI infrastructure stocks tumbled over 20%, resulting in a $600 billion decline in Nvidia's market value on Monday.

In response, Nvidia sought to contextualize DeepSeek's achievement. An Nvidia spokesperson told CNBC: "DeepSeek's work illustrates how new models can be created using test time scaling, leveraging widely-available models and compute that is fully export control compliant." The company further noted that DeepSeek's "inference requires significant numbers of Nvidia GPUs and high-performance networking," according to Reuters. Nvidia emphasized that DeepSeek used approved GPU versions designed for the Chinese market, countering claims about potential export violations.

The development reflects broader industry discussions about test-time scaling—a concept championed by Nvidia CEO Jensen Huang and OpenAI CEO Sam Altman. This approach suggests AI models can achieve better results by using additional computing power during inference rather than just during training. The achievement has sparked growing concerns about AI infrastructure spending, particularly given that DeepSeek demonstrated comparable performance to U.S. models at a fraction of the typical cost.

Nvidia's stock closed at $118.58 on Monday, down 16.86% for the day, though in after-hours trading the stock edged up 1.35%. Over the past year, Nvidia's stock surged 89.81%, according to data from Benzinga Pro. Despite the sharp single-day decline, the company's long-term performance remains significantly positive.