DeepSeek made a big impact not just on the Chinese AI market but also on the global sector. And as the Chinese startup gained attention, industry-leading experts were quickly asked to share their thoughts. Nvidia’s CEO, Jensen Huang, praised DeepSeek as a fantastic, world-class open-source model, while Google’s CEO called its work “very good“. The open-source nature of the AI model was a key talking point, with some experts even suggesting that DeepSeek could mark the end of closed-source AI.

Soon after DeepSeek R1 launched, AMD shared some promising RX 7900 XTX benchmarks. And now, in a recent interview with General Manager of ASUS China, Tony Yu, AMD CEO Lisa Su has shared her thoughts, calling DeepSeek a “very innovative model” and once again emphasizing that “it is an example of why open source is so important.”

Lisa Su says DeepSeek’s low computational requirements surprised the world

This is something we’ve been hearing since day one, as part of what made DeepSeek such a disruptive force in the AI industry was that it proved a sophisticated AI model could be created without relying on thousands of Nvidia GPUs. And a key example of its lower computational requirements comes from AMD itself, which shared DeepSeek's R1 AI inference benchmarks and demonstrated how the model could run on a consumer GPU like the RDNA 3-based RX 7900 XTX.

“I think DeepSeek is a very innovative model, and because of its open-source nature, it is an example of why open source is so important. The idea that you can accomplish such a sophisticated model with limited computing resources surprised the world.”

Dr Lisa Su, AMD CEO

When you put this into perspective, comparing it to other AI models, one example is Elon Musk’s latest Grok 3, which outperformed DeepSeek and showcased impressive capabilities but was trained on a supercomputer called Colossus that utilized about 200,000 H200 GPUs.

DeepSeek has sparked a faster pace of innovation

Lisa Su adds that “DeepSeek has also inspired an even faster pace of innovation than before,” which is something we continue to see as the race for the superior AI model drives tech giants like Alibaba and Google to release newer and improved versions of their AI models to compete with DeepSeek’s performance. Google’s Gemma 3 open-source AI model, in particular, ranks as the second-best open-model AI.

And competition isn't the only result of DeepSeek pushing companies into high gear, as collaboration is also taking shape. Nvidia has teamed up with DeepSeek for R1 optimizations on Blackwell, boosting revenue by 25x. And similar to Nvidia, AMD has also started to make a strong push into AI, which is seen not only in its latest RDNA 4 offering but also in its upcoming AI PCs. The company shared some interesting details about these during the ‘Advancing AI’ event in Beijing, which Lisa Su also discusses in this interview, explaining how AMD plans to make them the next evolution.

About the Author

Hassam Nasir

Hassam boasts over seven years of professional experience as a dedicated PC hardware reviewer and writer.