Baidu Launches "Project O": Evolving from Search to "Intelligent Agent Orchestration"

Deep News
4 hours ago

The atmosphere in China's internet sector is more intense than usual on the eve of the 2026 Spring Festival. On February 10, it was learned from internal sources at Baidu that the company has quietly initiated a project codenamed "Project O." Unlike previous departmental collaborations, this project involves the joint efforts of the search and cloud teams, signaling a breakthrough in departmental barriers with the Mobile Ecosystem Group and the AI Cloud Group working together. Other informed sources revealed that the collaborative project is related to the Baidu App. Baidu is targeting users' daily scenarios, aiming to use the Baidu App as a hub. Leveraging its ERNIE Assistant, the company intends to mobilize services from within its own ecosystem and those of partners to address users' practical needs. These actions send a strong signal. Baidu recognizes that the competition for AI access is no longer just about the front-end user interface; it has evolved into a comprehensive battle involving computing power costs, inference speed, and data orchestration capabilities. In simple terms, the UI is just a dialog box, something anyone can create. However, what truly creates a generational gap is the underlying orchestration logic and the accounting of computing power. When a user presents a complex request, the system needs to instantly mobilize services across the entire ecosystem. Poor orchestration capability reduces AI to a mere conversational ornament, unable to solve real problems. Conversely, low computational efficiency makes the cost per response prohibitively expensive, rendering the model commercially unviable. The launch of "Project O" was not without warning. In an internal sharing session early in 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 cannot significantly reduce the inference cost per intent resolution by deeply integrating its search and cloud infrastructure, then so-called "AI search" would become a loss-leading endeavor. Historically, the search team managed trillions of query flows, representing a cash cow and the company's existing business. In contrast, the large language model team represented the future and growth. Over the past two years, ERNIE Assistant and Baidu Search operated on parallel tracks. However, with the introduction of "Project O" in 2026, this separation has ended. Baidu Search is no longer just a display gallery; it has become the knowledge base for ERNIE Assistant. ERNIE Assistant, in turn, has taken the driver's seat. ERNIE Assistant acts as the front-end, responsible for perceiving user sentiment and ambiguous intent. Search serves as the back-end, tasked with extracting and verifying information from massive, real-time data. This model indicates that Baidu is no longer overly concerned with short-term fluctuations in search ad click-through rates. The rationale is that if Baidu does not revolutionize itself, competitors like ByteDance's Doubao and Alibaba's Tongyi Qianwen will challenge the traditional search ad model with their low-ad-interference, direct-result user experiences. The 2026 Spring Festival marks a "D-Day moment" for China's large model industry. As Baidu, Alibaba, Tencent, and ByteDance all plunge into the red envelope marketing battle, the conflict has transcended simple marketing tactics, evolving into a fight for survival centered on becoming the "default access point." In the competition for AI access, major tech companies are pursuing distinctly different paths based on their core strengths. As the official partner for the Spring Festival Gala, Doubao's strategy relies on its incredibly powerful content distribution network. ByteDance doesn't instruct users on how to think; instead, it uses short videos, social networks, and极致 UI interaction to position its AI assistant as a new tool for killing time. Doubao's high Daily Active User count is its core weapon, attempting to integrate AI into daily life through high-frequency engagement overwhelming low-frequency use. Tongyi Qianwen's massive 3 billion RMB "Spring Festival Treat" campaign makes its intention clear: AI must facilitate transactions. Alibaba is deeply integrating AI with local services and e-commerce, enabling actions like ordering milk tea or initiating returns with simple voice commands. Alibaba's AI access point is essentially a universal remote control for its vast commercial empire. Yuanbao, backed by WeChat, leverages close social relationships. It is not in a rush to operate independently but rather functions as an intelligent upgrade within the WeChat ecosystem, using social sharing of red envelopes to embed AI as a tool for social interaction. Now, Baidu has found its own positioning. It aims to solve users' specific, even complex, problems. Its reliance is likely the deep content accumulated through search and its full-stack AI capabilities. When ERNIE Assistant handles intent planning, Search provides knowledge verification, and Baidu AI Cloud ensures low-cost, high-concurrency, rapid inference execution, the blueprint for an AI agent orchestration center begins to emerge. By 2026, the so-called "Hundred-Model War" has completely faded. The industry has entered an intensely brutal period of battling for user access points. Baidu, leveraging its search scenario, holds a natural advantage in the AI access competition, but it inevitably faces the challenge of user diversion to standalone apps. Search is inherently the most natural ground for Q&A. If "Project O" can successfully transform search into an intelligent agent distribution platform, Baidu could potentially transition into an operator for the AI era. In the past, discussions about access points focused on traffic. Now, the conversation is about mindshare. Whoever can make users' first instinct, upon having a need, to speak directly to their phone instead of opening an app to search, will hold the pricing power for the future. Baidu's "Project O" is a high-stakes gamble, betting that users will develop a loyalty to receiving in-depth answers. Meanwhile, ByteDance and Alibaba are betting on instant gratification and convenience. This battle of differentiated strategies will be decided during the Spring Festival holiday, in every red envelope exchanged and every voice interaction. This is not merely a competition of technology; it is a game of courage. At the gateway to AI, there is no room for retreat.

Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.

Most Discussed

  1. 1
     
     
     
     
  2. 2
     
     
     
     
  3. 3
     
     
     
     
  4. 4
     
     
     
     
  5. 5
     
     
     
     
  6. 6
     
     
     
     
  7. 7
     
     
     
     
  8. 8
     
     
     
     
  9. 9
     
     
     
     
  10. 10