JAKARTA – Tech giant Nvidia has praised the innovation in Artificial Intelligence (AI) created by DeepSeek, acknowledging that it is likely to have a significant impact on the future of AI technology.

In an official statement, Nvidia noted that DeepSeek is an example of how new AI models can be developed by leveraging a wide array of existing models.

However, Nvidia pointed out that DeepSeek’s development still requires Nvidia’s chips. “The inference developed by DeepSeek relies heavily on Nvidia GPUs and high-performance networks,” the company stated.

In its research report, DeepSeek revealed that its AI technology is powered by around 2,000 H800 chips from Nvidia, a figure adjusted to comply with the export regulations set by the United States in 2024.

Furthermore, Nvidia emphasized that DeepSeek will need even more Nvidia chips in the future to carry out the Test Time Scaling phase. This phase refers to the optimisation of AI model performance during operation, typically occurring after Pre-Training Scaling and Post-Training Scaling have been completed.

It is worth noting that the emergence of DeepSeek has raised concerns among some that the technology industry may no longer require expensive AI infrastructure, following DeepSeek’s success in developing an AI model with more efficient and cost-effective resources. (KR/ZH)