“Artificial intelligence is not only a technological revolution but also an organizational revolution, and this trend is unstoppable.” At the opening ceremony of Baidu World 2026, Baidu founder Li Yanhong focused on the downstream of the AI technology stack: organizational capability. He believes that the next chapter of artificial intelligence is intelligent organizations, and companies need to redesign their organizations just as they would redesign a factory. He introduced for the first time the “Three Principles of Organizational Evolution”: breaking down job boundaries, consolidating tacit knowledge, and designing closed-loop processes. Li Yanhong pointed out that many companies are experiencing increasing Token costs, yet their revenues and profits have not grown correspondingly, because they have only replaced the motors without redesigning the factory.
In his speech, he shared Baidu's practices in internalizing AI capabilities, elaborated on the thoughts behind the large model trading platform Token Exchange, Kunlun chips, and autonomous driving, and proposed that intelligent organizations will encounter three major opportunities: light asset operations, building new moats, and creating and discovering new markets.
Intelligent Organizations as the New Factories in the AI Era
Looking back from 2016 to 2026, Li Yanhong stated that this decade has been the time when artificial intelligence technology has formed an independent technology stack, with the “chip-cloud-model” four-layer architecture gradually taking shape. He pointed out, “If the upstream of AI technology is energy, then the downstream of AI technology is organizational capability, which is a different AI-native organizational system from the current hierarchical structure, which we call intelligent organizations.” He predicts that the next chapter of artificial intelligence is intelligent organizations. He believes: “If artificial intelligence is the new electric power, then intelligent organizations are the new factories in the AI era, where productivity truly occurs.
Only when organizational methods that match various new capabilities of AI are implemented will productivity significantly increase, and the economy can accelerate its growth!”
Li Yanhong further stated that companies should not view AI merely as a tool for enhancing individual efficiency, but should integrate intelligent agents and digital employees into business processes such as sales, customer service, research and development, and operations, redesigning the organization around AI. He believes that many executives' questions about “why Token costs are increasing without a corresponding rise in revenue” are actually because they have only replaced the motors without redesigning the factory. “No company will become more competitive overnight just by using a few AI tools.”
Three Principles of Organizational Evolution
At this conference, Li Yanhong proposed the “Three Principles of Organizational Evolution.” The first principle is to break down job boundaries and unleash organizational vitality. He pointed out that jobs are units of the industrial era, while tasks are units of the intelligent era; the future organization will be a network of tasks. He believes that jobs will not disappear but will converge into two categories: Builders, who are responsible for creating new things, and Sellers, who are responsible for communication between people. “In the future, organizations will still exist, but they will be task-centric, forming a task network for collaboration between people and agents.”
The second principle is to consolidate tacit knowledge and build organizational intelligence. Li Yanhong stated that the biggest difference between organizational intelligence and individual capability lies in the explicitness of tacit knowledge. “True organizational intelligence is about making this knowledge explicit, consolidating it onto the company’s platform, allowing it to be continuously accumulated, readily accessible, and usable by everyone. Organizational intelligence is the deepest moat for a company.” The third principle is to design closed-loop processes to promote organizational evolution. He proposed the formula “measurable + Looping = continuous evolution.” He stated: “In the future, humans and AI will form a Loop around a task, and the iterative process we call Looping.
”
Li Yanhong further pointed out that the demand for model invocation is experiencing a massive explosion, and Token Exchange is a trading platform for large models. With the acceleration of model iteration speed, nearly every day in the first nine months of this year, new models have been released, and the demand for model invocation has also surged. To address this implementation challenge, Baidu officially launched the large model trading platform Token Exchange, aiming to gather outstanding models and model providers, allowing businesses and individuals to easily find, purchase, and use reliable model Tokens.
Inference Becomes a Key Driver for AI Chip Growth
Li Yanhong believes that inference will become the main driver for the long-term growth of AI chips. He stated: “In the past, the demand for AI computing power was mainly focused on training, but now there is a balanced emphasis on both training and inference, especially with the scaling of intelligent applications driving a massive surge in inference computing demand.” He pointed out that Baidu has always been a company that believes in technology and is willing to invest long-term, and Kunlun chips are the result of more than a decade of persistent investment.
Li Yanhong pointed out that autonomous driving will become an indispensable part of competition in AI technology. He believes that only by mastering all four-layer architecture technologies and ensuring compatibility between layers can the most valuable products be created. Baidu possesses a leading technological advantage in autonomous driving, and the development prospects for autonomous driving are promising.
Li Yanhong emphasized that intelligent organizations will encounter three major opportunities: light asset operations, building new moats, and creating and discovering new markets. He believes that with the emergence of intelligent organizations, human resources will no longer be a bottleneck for corporate expansion, and corporate innovation will become more flexible, thus increasing the likelihood of discovering and creating new markets.
Item Specification AI Technology Stack Four-layer architecture Autonomous Driving Self-Driving Service Model Globally unique Self-Driving Mileage Over 350 million kilometers

