Key Takeaways:
Baidu is shifting from a model-centric era to one centered on applications and agents, a move that could solidify its leadership in China's AI sector.
Baidu's stock surged over 5% after the company announced a strategic shift to an 'Agentic AI' era and launched new AI agents, 'DuMate' and 'Miaoda', positioning itself as a leader in the emerging agentic AI space.
"The Agentic AI era has arrived," Baidu Chairman and CEO Robin Li said at the Baidu Create 2026 conference, introducing "Daily Active Agents" (DAAs) as a new value metric to better reflect AI productivity output.
The company's stock (BIDU-SW) opened 6.98% higher, reaching a peak of HKD147.9. It closed at HKD145.1, up 5.53%, with a turnover of 2.6527 million shares. In response, Citi reiterated its Buy rating on Baidu, setting a US stock price target of USD186.
The move into agentic AI, which focuses on models that can autonomously perform multi-step tasks, could open significant new revenue streams for Baidu. This strategic shift intensifies the global "agentic wars," pitting Baidu against US giants like Meta and Google, who are also heavily investing in agent-based systems.
At its Create 2026 conference, Baidu launched two flagship agents: DuMate, a general-purpose intelligent agent, and Miaoda, a coding-specific agent. These products represent Baidu's effort to translate its foundational model capabilities into practical applications that can automate complex digital tasks, from managing spreadsheets to autonomously browsing the web for information.
This application-focused strategy is built on the foundation of Baidu's increasingly efficient models. The recently released ERNIE 5.1 cost 94% less to pre-train than comparable models by using a technique called "multi-dimensional elastic pre-training" to compress its predecessor, ERNIE 5.0. Despite having one-third the parameters, ERNIE 5.1 ranks fourth globally on the LMArena Search leaderboard, demonstrating that Chinese labs are making significant strides in reducing the cost of AI development without sacrificing performance.
Baidu's agent-centric push comes as the entire industry pivots from simply building large models to deploying them as active tools. This echoes the disruption caused by Chinese startup DeepSeek, whose low-cost inference models forced a re-evaluation of the "compute-is-king" mantra. With Meta and Google also entering the agent race, Baidu's introduction of "Daily Active Agents" as a key performance indicator suggests a new front is opening in the AI competition, focused on user engagement and task completion rather than just model benchmarks.
Citi's bullish stance reflects confidence in Baidu's ability to monetize this new strategy. The bank's research report highlighted Baidu's full-stack AI system—from its proprietary Kunlunxin chips to its AI cloud infrastructure—as a significant competitive moat. For investors, Baidu's HKD389 million in turnover following the announcement indicates strong market approval. The company's ability to control the entire stack, from hardware to application, could provide a crucial advantage in the capital-intensive agentic era, potentially shielding it from the margin pressures seen in pure model providers.
This article is for informational purposes only and does not constitute investment advice.
