Robin Li Says Entrepreneurs Competing in Large Models Is Meaningless: Greater Opportunities Lie in Application Development
-
At the Baidu 'Wenxin Cup' Entrepreneurship Competition awards ceremony on September 5, Robin Li, founder, chairman, and CEO of Baidu, emphasized that AI-native applications should be capable of solving problems that were previously unsolvable or poorly addressed, rather than simply replicating mobile internet apps or PC software.
He pointed out that while China has already released over 70 large-scale models with more than 1 billion parameters, these models themselves do not directly create value. Instead, it is the applications developed based on these foundational models that hold significance. Therefore, for entrepreneurs, competing in large models is meaningless—greater opportunities lie in developing applications.
Robin Li stated that AI-native applications should meet three fundamental criteria: First, they should be able to interact using natural language, which represents a fundamental shift. Second, they should fully leverage the capabilities of large language models, such as understanding, generation, reasoning, and memory—abilities that AI-native applications should harness. Third, the interaction for each application should not exceed two levels of menus to ensure simplicity.
He emphasized that only when large models generate a sufficient number of AI-native applications can a healthy ecosystem be formed. In the mobile internet era, there were only two operating systems—Android and iOS—but many successful applications emerged. Similarly, only when enough AI-native applications are built on large models can a thriving ecosystem be established.