Baidu closed 2025 with results that underline the perils of a company mid‑transition: full‑year revenue fell to RMB1,291bn, down 3% year‑on‑year, while attributable net profit plunged 76% to RMB56bn after a one‑off RMB162bn impairment charge. The impairment — a decisive write‑down of legacy computing infrastructure judged inadequate for modern AI workloads — turned a single quarter into a loss and shrank annual profits by roughly RMB18.2bn compared with the prior year.
The figures mask a more complex internal story. Traditional search and advertising, once Baidu’s cash engine, are suffering structural decline as users increasingly obtain information through social platforms and AI interfaces rather than classic blue‑link results. QuestMobile data cited in Baidu’s reporting show Baidu’s media status index slipping to ninth place by mid‑2025, trailing the likes of Douyin, Taobao, WeChat and Xiaohongshu.
Baidu’s corporate answer has been blunt: “All in AI.” Management has reorganised reporting to foreground an "AI new business" segment — comprising cloud infrastructure, AI applications and AI‑native marketing — and to separate it from a shrinking traditional business. That new segment delivered RMB400bn in 2025 revenue, up 48% year‑on‑year, and accounted for roughly 43% of core revenues in the fourth quarter, signalling that the pivot is producing growth, if not yet profit parity.
Yet the growth comes at a high cost. Baidu has invested more than RMB100bn in AI since 2023 and ran down older compute assets with the RMB162bn impairment, a move management framed as necessary to free capital and efficiency for frontier AI compute. The company also cut about 3,100 staff in late 2025 — predominantly from ad‑centric mobile ecosystems — and paid roughly RMB700m in severance to reduce headcount to about 29,000.
Despite signs of stabilisation — fourth‑quarter revenue of RMB327.4bn beat expectations and both operating and free cash flow turned positive — market reaction was sharp: Baidu’s U.S. stock tumbled as much as 7% on the results. Analysts and investors remain sceptical that AI revenue growth can fill the gap left by declining search ad sales, particularly because Baidu has not disclosed profitability metrics such as gross or net margins for its AI new business.
Competition compounds the challenge. A wave of AI product launches from ByteDance, Alibaba, Tencent and a host of smaller players has intensified the race for consumer attention. In user‑facing rankings of AI native apps, Baidu’s consumer products do not appear among the top ten, while rivals’ apps such as Doubao, Qianwen and Yuanbao have surged in downloads and active users through aggressive promotions and feature rollouts.
Operationally, the pivot has reached the CEO’s desk. Robin Li has tightened control over model development, creating two model R&D divisions reporting directly to him and absorbing parts of the CTO’s remit. The move signals a founder’s willingness to take personal responsibility for a make‑or‑break strategic bet, but it also raises governance questions about centralised decision making at a time when product execution speed matters.
Baidu’s short‑term outlook is therefore binary. If AI new businesses can improve monetisation, margins and consumer traction, the company can offset structural ad declines and benefit from first‑mover advantages in certain AI infrastructure stacks. If competitors continue to out‑execute on consumer apps and monetisation remains elusive, Baidu risks being a deep‑pocketed but strategically constrained incumbent, swapping one declining cash cow for an expensive, uncertain new core.