By the close of 2025, Baidu quietly initiated what the outside world is dubbing the “largest-ever” internal layoffs.

According to Reuters, starting from late November, Baidu embarked on structural adjustments involving multiple business lines, with layoff ratios generally ranging from 10% to 25%. Some non-core departments, such as live streaming and gaming, even witnessed layoffs as high as 90%. The offer was straightforward: sign and leave, with an N+3.5 compensation package, and both new hires and seasoned veterans were exiting simultaneously...

These layoffs represent more than just a simple downsizing; they are a strategic reset accompanied by significant pain.

Behind the generous compensation lies both a dignified farewell and Baidu’s intention to swiftly shed historical burdens and move forward with agility. The logic of “streamlining for efficiency” on the surface (this phrase is retained in Chinese pinyin as it is part of the original expression, meaning “on the surface”) suggests a focus on efficiency, but it actually reflects organizational uncertainty about the future. Against a backdrop of slowing growth momentum, companies often choose to cut burdens first to secure limited room for maneuver.

Advertising Business: The Stumbling “Cash Cow”

The direct catalyst for the layoffs is the significant decline in performance.

Once upon a time, “Baidu it, and you’ll know” was not just an advertising slogan but the core logic of China’s internet traffic distribution. Leveraging its absolute dominance in the search market, Baidu built an empire of “cash cows” centered around search advertising. However, today, this once-steady “old cow” is struggling, dragging down overall performance.

Financial reports reveal that Baidu’s revenue in the third quarter of 2025 was 31.2 billion yuan, down 7% year-on-year, marking the largest decline in history. More notably, there was a net loss of 11.2 billion yuan, a stark contrast to the 7.6 billion yuan profit in the same period last year. Online marketing revenue, the mainstay of revenue, was only 15.3 billion yuan, down 18% year-on-year.

In fact, since 2024, advertising revenue has been declining for multiple quarters, and its proportion of total revenue has also been decreasing since 2016. Financial reports show that in the first quarter of 2016, Baidu’s advertising revenue accounted for 94% of total revenue, dropping to 82% in the first quarter of 2018, 73% in the first quarter of 2019, and 58% in 2021. Now, it accounts for less than half.

The foundation of Baidu’s advertising business stems from its irreplaceable position as the “first entry point” for information. However, this status is being gradually eroded by the behavioral habits of a new generation of users.

First, advertising budgets follow traffic migration, with users shifting from “Baidu it” to platforms like Xiaohongshu, Douyin, and WeChat Search. QuestMobile data shows that in June 2025, Baidu’s media status index had fallen to ninth in the industry, behind Douyin, WeChat, and Taobao.

Second, macroeconomic and industry demand weakness has exacerbated the contraction in advertising spending. Against a backdrop of increased economic uncertainty, advertisers’ budgets have become more cautious, with significant cuts in spending on non-essential consumption, directly impacting Baidu’s revenue.

Furthermore, although Baidu has attempted to transition to AI-powered search to enhance monetization efficiency, new product forms and business models have not yet matured, making it difficult to offset the decline in traditional advertising in the short term.

In summary, the waning of Baidu’s advertising business marks the gradual retreat of the PC-era “entry logic”-based advertising model amid the intertwined waves of mobile and AI. Whether Baidu can establish a new foothold in the AI era will determine its future direction.

AI Business: Early Bird but Late Arrival

Among Chinese tech companies, Baidu has always carried the label of a “technological idealist.”

As early as 2013, when peers were still chasing the traffic dividends of the mobile internet, Baidu quietly established the Institute of Deep Learning (IDL), betting on a technological revolution with no clear commercial path. Over the next decade, from Baidu Brain and Apollo autonomous driving to Kunlun chips and Wenxin large models, Baidu has hit almost every key node in AI development, setting an example of being an “early bird.”

Today, Baidu’s AI layout is showing initial success.

In the third quarter of 2025, Baidu disclosed its AI business revenue for the first time, showing a year-on-year increase of over 50%. Among them, AI cloud revenue grew 33% year-on-year; AI-native marketing service revenue surged 262% year-on-year to 2.8 billion yuan; AI application revenue reached 2.6 billion yuan, covering flagship products like Baidu Wenku, Baidu Netdisk, and digital employees.

Notably, Luobo Kuaipao’s global ride service volume reached 3.1 million in the third quarter, up 212% year-on-year, accelerating from the 148% growth in the second quarter. In October, Luobo Kuaipao’s weekly fully autonomous ride volume exceeded 250,000; as of November, the cumulative global ride service volume surpassed 17 million.

However, behind the impressive data, Baidu’s AI shortcomings are equally hard to ignore: deep technological accumulation but slow product landing, significant first-mover advantage but weak market presence.

Take large models as an example. Wenxin 1.0 was released in 2019, long before ChatGPT took the world by storm. However, Baidu still viewed it as internal technological infrastructure at the time, lacking a product-oriented mindset for consumers or businesses. It wasn’t until the launch of Wenxin Yiyan in 2023 that Baidu truly attempted commercialization, missing the best window to capture user mindshare.

In contrast, competitors like ByteDance, although starting later, quickly integrated “Doubao” as a “life assistant + office efficiency booster” into ecosystems like Douyin and Feishu, seeing rapid growth in monthly active users. Alibaba’s “Qianwen” also leveraged the Qwen3-Max large model and ecosystem synergy to show strong growth momentum.

QuestMobile’s “2025 Q3 AI Application Value Rankings” show that ByteDance’s Doubao has 159 million monthly active users and 34.47 million downloads; Baidu Wenxiaoyan has only 5.167 million monthly active users. Alibaba’s Qianwen surpassed 30 million monthly active users in just 23 days, becoming one of the fastest-growing AI applications globally.

Faced with saturated investments and rapid iterations from ByteDance, Tencent, Alibaba, and others, Baidu’s products are trapped in a dilemma of technological leadership but lagging user experience, rooted perhaps in a lack of strategic focus.

This lack of focus is first reflected in the constant oscillation of brands and products. Over the past two years, Baidu’s large model names have cycled through “Wenxin Yiyan, Wenxiaoyan, Wenxin,” diluting brand recognition through repeated adjustments.

Deeper instability comes from ongoing organizational restructuring. On November 25, Baidu announced the establishment of the Foundational Model R&D Department and Applied Model R&D Department, both reporting directly to CEO Li Yanhong. Wang Haifeng, the group’s CTO, was excluded from core R&D management for large models, reflecting a high-level reassertion of control over technological landing pace.

The most critical shortcoming may lie in the disconnection of ecosystem synergy. Alibaba has e-commerce and cloud ecosystems, Tencent has social and content systems, and ByteDance dominates short video and advertising scenes. Their AI capabilities naturally integrate into high-frequency businesses, forming closed loops of “technology-scene-data-feedback.” Baidu, although possessing entry points like search, netdisk, and maps, has failed to achieve effective integration and lacks deep consolidation.

Next, Baidu’s real test lies in whether it can drive the AI business toward scalable profitability before cash flow pressures intensify further.

A Brutal Prelude to a Hundred-Billion AI Gamble

For Baidu, artificial intelligence is undoubtedly a strategic pathway to the future. Bain & Company’s “2024 Global High-Tech Industry Report” predicts that the AI software and hardware market will continue to expand at an annual growth rate of 40% to 55%, reaching an estimated market size of $780 billion to $990 billion by 2027.

However, on one side is the bleeding of traditional businesses and layoffs, and on the other is continuous investment in AI. In this arduous transformation from a “search giant” to an “AI company,” Baidu is experiencing intense internal and external pain.

First, this gamble comes with unprecedented competitive pressure. The current AI track is already a red ocean, with domestic giants like Huawei Pangu, Alibaba Tongyi, Tencent Hunyuan, ByteDance Doubao, and iFLYTEK Xinghuo vying for dominance, and the arms race intensifying. More seriously, international top players like OpenAI and Anthropic continue to widen the technological gap with stronger computing power and ecological advantages, leaving narrower time windows for latecomers.

Second, shifting from traffic-driven to technology-driven means reconstructing organizational DNA. This involves systemic changes in talent structure, incentive mechanisms, and decision-making processes. However, Baidu’s ongoing organizational turmoil and strategic oscillations have objectively weakened internal morale and external confidence.

If consensus cannot be truly forged and the course stabilized, today’s “largest-ever layoffs” may become the prelude to an even greater crisis in the future.

In conclusion, Li Yanhong’s AI gamble is both a farewell to the past and a bet on the future. Success means rebirth; failure means accelerated marginalization. 2025 may be Baidu’s most painful year, but it could also be the starting point for renewal. As 2026 approaches, is Baidu ready?