On January 20, Baidu announced that its Wenxin Assistant has surpassed 200 million monthly active users (MAU), joining ByteDance's Doubao and Alibaba's Tongyi Qianwen as the "Big Three AI Giants."
The announcement was immediately met with widespread skepticism: "Haven't they already fallen behind?" "How was this data calculated?" "Does forcing it through search count as real users?"
On the surface, this appears to be the final outcome of the "Hundred-Model War"—a battle among over a hundred large models has narrowed down to a tripartite standoff. However, looking past the data fog reveals a more complex truth: Baidu hasn't staged a comeback; instead, it has transformed its "gateway" into "infrastructure."
Many forget that Baidu still handles over 600 million search queries daily.
When a user asks "what's good to eat nearby" within the Baidu App, the response is no longer a list of links. Instead, Wenxin Assistant directly interfaces with Meituan, Ctrip, and JD.com to generate a complete solution with booking buttons.
When a student searches for "quantum mechanics for beginners," Wenxin can not only explain the concepts but also link to Baidu Wenku for textbooks and use AI to generate a study plan. This isn't a user acquisition miracle for a standalone app, but rather a "silent conversion" achieved by embedding AI into China's largest information gateway.
Users don't download new software; they might not even realize "I'm using AI." Yet, every search interaction is counted towards the monthly active user tally.
This strategy seems clever, but it's actually shrewd—it avoids the bloody competition for users head-on with Doubao and Tongyi Qianwen, and instead focuses on cultivating the "first point of demand."
You don't need to actively seek out AI; AI appears the moment you ask a question. The 200 million MAU wasn't achieved through massive subsidies, but through the natural accumulation of 600 million daily search visits.
Why has it become a "Tripartite Standoff"? Because the battlefield has changed.
The early "Hundred-Model War" was about comparing parameters, open-source contributions, and technical benchmarks. Today, the decisive factor has shifted to "who can truly serve high-frequency scenarios."
ByteDance leverages the Douyin content ecosystem to position Doubao as a "creation companion"; Alibaba uses Taobao and Alipay to turn Tongyi Qianwen into a "life manager"; and Baidu is betting on the most fundamental and enduring need—information acquisition.
In this dimension, Baidu holds a natural advantage: - Users arrive with clear intent ("I want to know..."), not for casual chat; - The search scenario inherently requires multi-tool integration (lookup, booking, buying, learning); - Baidu's own products like Wenku, Wangpan, and Maps form a closed-loop service chain.
Thus, Wenxin Assistant doesn't need to show off; it simply needs to silently call Ctrip's API when a user says "help me book a flight to Shanghai for tomorrow"—utility is the best retention strategy.
Has it truly staged a comeback? Don't be fooled by the data.
Looking at it calmly, Baidu's "200 million" still harbors underlying concerns.
First, the quality of engagement is questionable. AI invocations in search scenarios are mostly single-use and low-frequency, far from Doubao's average of 5 daily dialogues per user or Tongyi Qianwen's deep task processing. Users might leave immediately after use, with no stickiness.
Second, there is insufficient ecosystem synergy. Alibaba has e-commerce, finance, and logistics; ByteDance has content, social media, and e-commerce; whereas Baidu, beyond search, lacks strong transactional scenarios. Wenxin can "talk" and "look up," but struggles to "close the deal," leaving its commercial monetization path unclear.
More critically, its technical reputation hasn't fully recovered. Although ERNIE 5.0 ranked second globally in mathematical reasoning, this has weak perception among ordinary users. They only remember "Wenxin's answers feel like ads" or "it's still not easy to use even after three rebrandings." Once trust is damaged, rebuilding it takes years.
So, the tripartite standoff is real, but Baidu's leg of the tripod still seems somewhat unsteady.
The battle over gateways is essentially a battle for "scenario sovereignty."
This evolving landscape reveals a harsh reality: large models themselves have become a commodity. Victory no longer hinges on technical superiority, but on who controls the "scenario sovereignty"—the contexts where users most frequently use AI.
ByteDance wins in the "content creation scenario," Alibaba prevails in the "life service scenario," and Baidu holds the "information query scenario"—for now, the three parties each hold their corner in a temporary equilibrium.
However, variables for the future remain. If WeChat fully opens up AI search, or if Douyin successfully creates a closed loop for local services, Baidu's moat may not be unbreachable.
200 million is a starting point, not the finish line.
Baidu's 200 million MAU is neither a myth nor a scam, but a phased achievement of a pragmatic strategy. It hasn't raced ahead under the spotlight, but has quietly taken root within the search box.
Yet the real test is just beginning: How to make users transition from "passive use" to "actively seeking it out"? How to evolve from an "information assistant" to an "action agent"? How to transform its advertising DNA into service value without causing反感?
The tripartite standoff is merely the halftime whistle. The next phase of the competition will not be about who is the loudest, but about who can truly integrate AI into the real lives of hundreds of millions of ordinary people—
And not just leave it as a pop-up after a single search.
Comments