Nvidia has recognized the significance of DeepSeek’s R1 reasoning model, calling it an “excellent AI advancement.”

This acknowledgment comes as the Chinese startup’s emergence contributed to a 17% drop in Nvidia’s stock price on Jan. 27.

DeepSeek’s R1, an open-source reasoning model, reportedly outperformed US counterparts like OpenAI at a fraction of the training cost, less than US$6 million compared to billions spent by Silicon Valley firms.

Nvidia emphasized that DeepSeek’s progress demonstrates demand for its GPUs, which remain export compliant despite claims of unauthorized use.

The company highlighted “test-time scaling,” a concept enhancing AI model performance during inference by leveraging additional computing power, as pivotal to advancements like R1.