Nvidia calls DeepSeek’s R1 model 'an excellent AI advancement'

By Gokhan Ergocun

ISTANBUL (AA) - US chip giant Nvidia described the R1 model launched by Chinese artificial intelligence (AI) startup DeepSeek as “an excellent AI advancement” Monday, while noting that a significant number of chips will be needed to meet the demand for its services.

The Nvidia's spokesperson’s comments came after DeepSeek released R1 last week, a free AI assistant that it says uses much less data than its rivals at a fraction of the cost, triggering a major tech stock selloff Monday and causing a huge loss in the value of shares of chip companies, including Nvidia.

DeepSeek’s work illustrates how new models can be created using the technique, leveraging widely-available models and compute that is fully export control compliant, the statement said, adding the model is "an excellent AI advancement and a perfect example of Test Time Scaling," Forbes reported.

Inference, the process of providing a trained model inputs to generate outputs such as images, text, or video and make accurate predictions, "requires significant numbers of NVIDIA GPUs and high-performance networking," the statement underlined.

"We now have three scaling laws: pre-training and post-training, which continue, and new test-time scaling."

The latest Chinese DeepSeek AI model uses fewer chips than international competitors.

The model became the most downloaded AI application in app stores, surpassing US-based ChatGPT.

The share price of Nvidia, which was $142.62 at the close on Friday, had fallen around 17% to $118.19 as of 2150GMT.

While Nvidia's shares have seen the highest daily loss since March 2020, the company's market capitalization has also decreased by around $500 billion.