Baidu’s AI Transition Is Real—but Q2 Still Supports a HOLD on BIDU

Baidu is no longer primarily an internet-search story. Its AI-powered businesses reached RMB12.5 billion in Q2 2026, equal to 50% of Baidu General Business revenue and up 25% year over year. But the transition is not yet complete: total revenue declined, online marketing weakened sharply, and the company still faces intense competition in AI infrastructure and uncertainty around autonomous driving. 21416

The key clarification: AI accounts for half of General Business, not total company revenue

The headline that AI contributes more than half of Baidu’s revenue needs a precise definition. In Q2, Baidu Core AI-powered Business generated RMB12.5 billion, exactly 50% of Baidu General Business revenue of RMB25.2 billion. 214

Baidu’s consolidated revenue—including iQIYI—was RMB31.3 billion. On that broader basis, the AI-powered figure represented roughly 40% of total company revenue, not more than half. 216 The strategic milestone is still significant: AI has become central to Baidu’s core business, but the company has not yet replaced its legacy revenue base.

AI cloud is the clearest growth engine

AI Cloud Infrastructure revenue rose 50% year over year to RMB7.3 billion in Q2, while GPU-cloud revenue increased 283%. 24 These figures show strong demand for computing capacity and give Baidu a rapidly growing business beyond traditional search advertising.

The more important investment question is whether that growth can become durable, profitable and defensible. Baidu competes with major Chinese technology platforms such as Alibaba Cloud, Tencent Cloud and Huawei in a market requiring substantial investment in GPUs, infrastructure, models and enterprise distribution. 11

That makes AI cloud a growth engine, but not automatically a moat. Fast revenue growth can coexist with pricing pressure, high capital requirements and uncertain long-term margins. Investors therefore need to watch profitability and customer retention—not just GPU growth rates.

Search advertising remains the transition’s central problem

Baidu’s online marketing revenue fell 19% year over year to RMB13.1 billion in Q2. Analysts attributed the pressure to weak advertising demand, while management also pointed to competition for user attention and a deliberate delay in monetizing AI search.

Generative-AI answers can change the economics of search by resolving queries directly rather than sending users through conventional result pages and advertising links. That creates a structural challenge for Baidu: its newer AI products must grow quickly enough to compensate for pressure on a historically important, high-margin business.

The Q2 numbers show that this replacement process is still unfinished. Consolidated revenue fell 4% year over year to RMB31.3 billion and came in below the cited analyst estimate, despite strong AI-cloud growth. 1

Apollo Go offers scale and long-term optionality

Apollo Go gives Baidu another potential growth avenue. By Q2, the robotaxi business was reported to operate in 28 cities worldwide, with more than 23 million cumulative rides and over 350 million autonomous kilometers. 5

Those figures demonstrate operating scale, but they do not by themselves establish attractive economics. Autonomous ride-hailing still requires fleet deployment, remote support, maintenance, safety compliance, city-by-city approvals and public acceptance. The available evidence supports viewing Apollo as valuable optionality rather than a proven near-term earnings driver.

Baidu has also cited a rate of one airbag deployment every 14.4 million kilometers. That is a company-reported safety measure, not an independently standardized comparison, so it should be treated as an operating indicator rather than conclusive proof of industry leadership.

Regulation and reliability remain material Apollo risks

Apollo’s expansion has already shown how operational incidents can become regulatory obstacles. After Apollo Go vehicles abruptly stopped in Wuhan and disrupted traffic, China reportedly suspended the issuance of new autonomous-vehicle licenses while authorities reviewed safety and monitoring procedures. 1718

Baidu later said domestic ride volumes were temporarily affected by operational adjustments related to regulatory considerations. Reports in July indicated that robotaxi license approvals had begun to resume gradually, but the episode illustrates the conditions attached to scaling: progress depends not only on technology, but also on reliable fleet operations and regulatory approval in each market.

Why the investment case remains HOLD

The bullish case is straightforward: Baidu is building a meaningful AI-cloud business, AI now represents half of General Business revenue, and Apollo could provide additional long-term value. 214

The bear case is equally clear. Search advertising is shrinking, AI search monetization remains deliberately restrained, infrastructure competition is intense, and Apollo’s commercialization is exposed to safety, regulatory and operational setbacks. 111618

The cited valuation analysis produced a DCF fair-value range of $90.25–$102.70, suggesting limited upside near the share price referenced after the Q2 results. That leaves little margin of safety while Baidu proves whether its AI growth can replace declining legacy revenue. 11

For that reason, HOLD is the more balanced conclusion as of August 26, 2026—not because Baidu lacks growth opportunities, but because the evidence has not yet shown that those opportunities will generate enough durable earnings to justify a stronger rating.

What could change the rating?

A more bullish view would require evidence across three areas:

• Search stabilization: AI-search engagement and monetization begin to offset the decline in conventional online marketing.
• Profitable cloud growth: AI Cloud Infrastructure maintains strong growth while demonstrating durable margins despite competition and investment needs.
• Reliable Apollo expansion: Robotaxi operations expand across cities without repeated disruptions, and regulatory approvals translate into a credible path toward attractive unit economics.

Until those signals emerge, Baidu’s transformation is best understood as promising but still in execution mode.