AI Computing Power Development Forum｜Li Zixue: DeepSeek's rise brings two new challenges for AI computing

[Photo/wicinternet.org]

The World Internet Conference hosted the AI Computing Power Development Forum themed on "Building an Integrated, Inclusive, and Green AI Computing Ecosystem" in Barcelona, Spain, on March 4, during the 2025 Mobile World Congress. 

ZTE Corporation Chairman Li Zixue, one of the speakers at the event, highlighted three key issues in the ongoing digital transformation: the increasing demand for computing power, the need for greater energy efficiency, and the rapid expansion of AI applications.

Li noted that the current shortage of computing power is driven by the growing demand for processing vast amounts of data generated by the Internet of Everything. The emergence of large AI models in recent years has further intensified this challenge.

The global popularity of DeepSeek earlier this year has shifted the computing bottleneck from high-performance training to the broader need for inference computing. According to Li, this presents two major challenges: the urgent need to expand inference computing capacity and the continued necessity of enhancing training power to support larger and more complex AI models.

Li also addressed concerns about the commercial viability of AI computing centers, which have seen rapid expansion in recent years. He noted that DeepSeek has achieved breakthrough in cost and efficiency, enabling faster adoption and deployment of AI computing solutions. “We are witnessing the next wave of AI application growth,” he said.

While AI has been widely applied in general fields, challenges such as usability and accessibility issues remain in industry-specific adoption, according to Li. To make AI more effective for businesses, he suggested tailored model training and deeper integration with enterprise workflows. He also emphasized the importance of localized deployment to ensure data security and proposed a step-by-step approach to digital transformation, starting with targeted applications before expanding across industries.

"As we move forward in the intelligent era, the ICT industry must tackle three key challenges: overcoming computing power constraints, advancing green technology, and expanding AI applications. These challenges are also opportunities,” Li concluded.