A restructuring that makes Baidu’s priorities unmistakable
Baidu has begun a broad round of layoffs across several business units following a loss-making third quarter, while pledging to preserve roles tied to AI and cloud computing. The move, now rolling through teams in Beijing, Shanghai, and Guangzhou, is expected to run through year-end and may reach deep into parts of Baidu’s mobile ecosystem. Yet the company’s internal message is clear: legacy internet operations will shrink, and capital as well as talent will be redirected to the businesses Baidu believes can still compound. In China’s intensifying AI race, this is not only cost control. It is a strategic reset that reveals how Baidu now sees its own future.
A Q3 loss driven by legacy pressure, not an AI collapse
Baidu’s third-quarter results showed the tension at the heart of its transition. Revenue fell about 7% year-on-year to roughly US$4.4 billion, marking one of its sharpest quarterly declines in years. The core online marketing business, still Baidu’s main cash engine, dropped about 18% as advertisers shifted budgets toward short-video and social platforms. The quarter closed with a net loss of about US$1.6 billion, largely because Baidu booked a sizeable impairment charge after reviewing older assets in its traditional internet stack.
At the same time, Baidu’s newer AI-powered businesses continued to grow. AI-related revenue jumped roughly 50% year-on-year, and AI Cloud posted strong double-digit expansion, supported by enterprise adoption of model APIs and compute leasing. Non-marketing revenue, driven mainly by cloud, rose about 21%. This is important context. Baidu is not cutting because AI failed. It is cutting because the legacy ecosystem can no longer subsidize the transition at the old scale.
What the layoffs target, and what they protect
The cuts are concentrated in units most exposed to the slowing ad economy. Internal estimates suggest that some teams may lose as much as 30–40% of headcount, with the mobile ecosystem and search-adjacent product groups taking the hardest hit. Baidu has already been shrinking in absolute size over recent years, with total employees down to about 35,900 in 2024 from more than 41,000 in 2022. This round accelerates that trend.
What Baidu is not cutting is equally telling. Roles linked to the company’s foundation-model program and to Baidu AI Cloud are broadly protected, and some teams are reportedly receiving more resources. Baidu’s public filings highlight AI Cloud as the core commercial bridge between its Ernie models and paying enterprise customers. The company is also still investing in AI infrastructure, including its in-house chips for training and inference, to reduce dependence on constrained external supply.
This “shrink here, double down there” posture reflects a very specific bet. Baidu believes that defensible growth will come from enterprise AI services and cloud-delivered intelligence, not from trying to revive a weakening ad-search flywheel. It is a pivot from audience scale to capability scale, and it requires a different cost structure.
Baidu is choosing to be an AI company, not an internet conglomerate
For most of its life, Baidu was built around search and traffic monetisation. That model is now structurally pressured in China. Discovery has moved to short video, lifestyle networks, and super-apps. Advertising budgets follow attention. In that environment, keeping a large mobile-internet perimeter alive would force Baidu into a slow bleed. The layoffs acknowledge that reality.
The harder question is whether AI and cloud can replace the lost cash flow quickly enough. Baidu was the first major Chinese tech firm to release a ChatGPT-style bot in 2023, yet it has struggled to maintain leadership in the consumer AI layer. Ernie’s monthly user base remains small relative to rival products from ByteDance and DeepSeek, and that gap weakens Baidu’s consumer-platform leverage. The company, therefore, is building a different AI identity: less “winner-takes-all chatbot,” more “enterprise AI backbone.”
This is a credible path. Enterprise AI spending is accelerating in China as banks, manufacturers, and public-sector agencies seek domestic model partners. Cloud-based model deployment also creates recurring revenue rather than one-off licensing. Still, the shift requires Baidu to execute at two levels simultaneously: deliver measurable ROI for enterprise clients, and keep AI cost curves manageable as compute demand rises.
A slimmer Baidu, with higher stakes in AI execution
Over the next few quarters, investors will watch for two proof points. The first is whether AI Cloud can sustain growth while overall revenue shrinks. Strong cloud momentum would validate Baidu’s move to protect AI roles and cut legacy headcount. Weak cloud numbers would raise concerns that Baidu is retrenching without a sufficiently strong replacement engine.
The second proof point is commercialisation of the Ernie ecosystem. Baidu does not need to win the entire consumer bot war to succeed, but it does need a large enough model footprint to power enterprise use cases, developer tooling, and vertical AI products. If Ernie-powered services become standard across regulated industries, Baidu can still secure a long-term moat even with a smaller consumer presence.
There is also a strategic spillover into autonomous driving. Baidu’s robotaxi platform Apollo Go continues to scale rides and refine fleet economics. A healthier AI core could strengthen this business, because autonomy depends on exactly the kind of model and cloud stack Baidu is trying to defend.
In short, Baidu is trading breadth for depth. It is stepping away from being a sprawling internet ecosystem and betting that a tighter AI-first portfolio can drive its next decade.
Cost cuts as a roadmap, not a retreat
Baidu’s layoffs are a painful move for employees, yet strategically coherent for a company navigating a tectonic shift. The third-quarter loss, driven by asset write-downs and a weakening ad core, forced Baidu to accelerate restructuring. By cutting legacy units while shielding AI and cloud teams, Baidu is making a high-conviction bet on where growth will come from in China’s next tech cycle. The question now is not whether Baidu can get leaner. It is whether its AI and cloud businesses can grow fast enough, and profitably enough, to justify the future it is choosing.
