TMTPOST -- Baidu CEO Robin Li emphasized the ongoing need for investment in AI infrastructure, stating that advancements like DeepSeek’s efficient models do not eliminate the demand for high-performance computing. Speaking at the World Government Summit in Dubai on Tuesday, Li highlighted the importance of cloud infrastructure and data centers in maintaining a competitive edge in AI. “The investment in cloud infrastructure is still very much required. To develop models that surpass others in intelligence, you must leverage more compute,” he told attendees. "Compute" refers to the processing power needed for AI models to analyze data, generate responses, and improve performance. Li’s remarks follow the rise of DeepSeek, a Chinese AI startup that has gained global recognition for producing language models with performance comparable to OpenAI’s GPT while significantly reducing computing costs. DeepSeek’s success has sparked debates over whether massive AI infrastructure investments are still necessary. Baidu was one of the first Chinese companies to enter the AI space following OpenAI’s release of ChatGPT in late 2022. Its flagship large language model, Ernie, is claimed to rival GPT-4 in capability but has struggled with widespread adoption. Li, who has previously argued that an OpenAI-like company was unlikely to emerge in China, acknowledged that DeepSeek’s rapid progress underscores the unpredictable nature of AI breakthroughs. “You never know when and where innovation will come from,” he said. He also noted that U.S. chip sanctions have pushed Chinese companies to innovate despite hardware limitations. In a notable shift, Li appeared more open to the potential of open-source AI, contrasting his previous stance that closed models were the only viable development path. “If you open things up, more people will be curious enough to try it. This will help accelerate AI adoption,” he concluded.