Search Restructuring.
Author | Zhou Zhiyu
On the eve of the Spring Festival in 2026, the atmosphere of competition in China's internet sector became more intense than in previous years.
On February 10, Wall Street News learned from insiders that Baidu had quietly launched a project codenamed 'O Plan' within the company recently. Unlike previous departmental collaborations, this initiative involves joint efforts between the search and cloud teams, indicating that the project breaks down departmental barriers, with MEG and ACG working together. $Baidu (BIDU.US)$ 、 $BIDU-SW (09888.HK)$
According to another source familiar with the matter, this collaboration is related to the Baidu App. Baidu is targeting users' daily scenarios, using the Baidu App as a hub to leverage Wenxin Assistant and mobilize services from both Baidu’s internal ecosystem and its partners to address users' practical needs.
These actions send a strong signal. Baidu has realized that the competition for AI entry points is no longer just about front-end UI but rather a comprehensive battle involving computing power costs, inference speed, and data scheduling capabilities.
To put it bluntly, the UI is merely a dialog box, which anyone can create; what truly sets generational differences apart is the underlying scheduling logic and computational cost accounting. When a user presents a complex need, the system must instantly mobilize services across the entire ecosystem. Poor scheduling capability would render AI an empty shell capable only of conversation, unable to solve real problems; while low computational efficiency would make the cost per response prohibitively expensive, rendering the business model unfeasible.
The launch of Baidu’s 'O Plan' was not without foreshadowing. In an internal sharing session at the beginning of 2026, Baidu founder Robin Li clearly stated that the purpose of training is inference.
This implies that inference efficiency will determine the generational competition in search. If Baidu fails to integrate search and cloud infrastructure to significantly reduce the inference cost of single-intent parsing, so-called 'AI search' would become a losing proposition.
In the past, the search team guarded trillions of query traffic, representing a cash cow and existing assets; whereas the large model team symbolized the future and growth potential. Over the past two years, Wenxin Assistant and Baidu Search operated under two parallel logics. However, in 2026, with the unveiling of 'O Plan,' this state came to an end.
Baidu Search is no longer a mere repository but serves as the knowledge base for Wenxin Assistant. Meanwhile, Wenxin Assistant has become the driver holding the steering wheel.
Wenxin Assistant acts as the front end, responsible for perceiving user emotions and vague intents; search serves as the back-end verification, responsible for extracting authenticity checks from vast amounts of real-time data.
This model implies that Baidu no longer fixates on short-term fluctuations in search ad click-through rates (CTR). Because if Baidu does not revolutionize itself, ByteDance's DouBao and Alibaba's Qwen will disrupt the traditional search ad model with an interaction experience characterized by 'minimal ad interference and direct delivery of results.'
The Spring Festival of 2026 marks the 'D-Day moment' for China’s large-scale AI model industry.
As Baidu, Alibaba, Tencent, and ByteDance collectively engage in the red envelope war, this battle has long surpassed a simple marketing strategy, evolving into a life-or-death struggle over the 'default entry point.'
In the competition for AI entry points, major tech companies have pursued vastly different paths based on their respective core competencies.
DouBao, as the partner of the Spring Festival Gala, operates on an extremely powerful content distribution logic. ByteDance does not teach users how to think; instead, it turns the AI assistant into a new toy for users to pass time through short videos, social chains, and ultimate UI interactions. DouBao's high DAU is its core weapon, attempting to make AI an integral part of life by leveraging high-frequency engagement to counter low-frequency use cases.
Qwen launched a 3 billion yuan 'Spring Festival treat,' revealing its clear intent—AI must drive transactions. Alibaba deeply integrates AI with local services and e-commerce, enabling tasks like ordering milk tea or returning goods with a single voice command. Essentially, Alibaba’s AI entry point functions as a universal remote control for its vast commercial empire.
Yuanbao, backed by WeChat, capitalizes on personal relationship networks. Yuanbao is not in a rush to become independent; rather, it serves as an intelligent upgrade within the WeChat ecosystem, embedding AI into social interactions by distributing red envelopes through social channels.
Now, Baidu has also found its positioning. It aims to address users’ specific and even complex queries, relying on its accumulated depth of content and full-stack AI capabilities from its search business.
When Wenxin Assistant handles intent planning, search manages knowledge validation, and Baidu Intelligent Cloud ensures cost-effective, high-concurrency rapid inference execution, the prototype of an AI agent dispatch center may begin to emerge.
By 2026, the so-called 'hundred-model war' had completely dissipated. The industry entered an extremely brutal phase of competing for entry points.
Baidu, relying on its search scenarios, has a natural advantage in the competition for AI entry points, but it inevitably also faces the challenge of user traffic being diverted to independent apps.
Search itself is the most natural ground for question-and-answer interactions. If the 'O Plan' can successfully transform search into a platform for distributing intelligent agents, Baidu could also transition into an operator for the AI era.
In the past, when discussing entry points, the focus was on traffic. Now, when discussing entry points, the emphasis is on mental engagement. Whoever can make users speak directly to their phones at the first thought of a need, instead of opening an app to search, will control the pricing power of the future.
Baidu's 'O Plan' is a bold gamble, betting that users will develop loyalty to deep, insightful answers. Meanwhile, ByteDance and Alibaba are banking on delivering instant gratification and convenience. This differentiated battle will reveal its results during every red packet exchange and every voice interaction over the Spring Festival.
This is not just a contest about technology; it is also a game of courage. At the gateway to AI, there is no turning back.
Editor/Doris
