Baidu heads into its September-quarter report with two businesses pulling in opposite directions. Its GPU cloud operation has accelerated for four straight quarters, while the advertising engine that once defined the company is shrinking by design as search is rebuilt around AI. Last quarter the second force won the headline numbers: revenue and margins fell sequentially and the stock has since lost more than 13%. This report is a test of whether cloud momentum is finally large enough to move the consolidated results, or whether it remains a compelling sideshow.
The Street expects non-GAAP earnings of $0.85 per share on revenue of about $4.61 billion. That revenue figure is essentially flat with the prior quarter's $4.62 billion but implies roughly 5% growth from a year ago. This would be a modest improvement after last quarter's 4% yuan-based decline. Some of the dollar-based gain could reflect currency translation, so the yuan figures will be the cleaner read. Earnings remain well below the $1.23 posted a year earlier, a reminder that Baidu is spending heavily through this transition. The earnings whisper of $0.91 sits meaningfully above consensus, suggesting the more engaged crowd sees room for a bottom-line beat off last quarter's depressed $0.75. Management does not issue formal numerical guidance, so its qualitative signals carry more weight than usual.
Those signals were clear. Executives pointed to a strong second half for AI cloud, with potential for further acceleration after GPU cloud revenue jumped 283% year over year, up from 184% the quarter before. AI Cloud Infra grew 50% with expanding margins, and external token revenue on the Qianfan platform rose more than ninefold. Validation this quarter would mean GPU cloud growth holding at or above that pace and AI Cloud Infra growth stabilizing after slowing from 79% to 50%. It would also mean the core AI-powered business moving back above half of General Business revenue after slipping last quarter. Evidence that the Kunlunxin M100 inference chip is contributing to demand would strengthen the case that Baidu owns a differentiated stack rather than renting out commodity compute.
The drags matter just as much. Management warned advertising would stay under pressure through the second half because it is deliberately deferring monetization of AI search. A further decline in ads is therefore expected. The real question is whether it is shallow enough for cloud to offset it. Profitability is the other swing factor. Non-GAAP net margin fell to 8% from 14%, hurt by weaker other income and foreign exchange losses, so the quality of any beat will matter. A beat driven by investment gains would say far less than one driven by operating leverage in General Business. ApolloGo is a third checkpoint. Fully driverless rides collapsed to about 1 million from 3.2 million after regulatory pauses in some domestic cities. Management said those cities began resuming in August, so a visible rebound in ride volume would confirm the setback was temporary, while another weak tally would raise harder questions about China's tightening autonomous driving rules.
Sentiment has cooled, slipping from mildly bullish before the last report to slightly bearish now. The stock has trailed the S&P 500 by nearly 9 points since the last report, and at about $82 it sits far below its 200-day average near $113. It trades closer to the bottom of its post-earnings range of roughly $76 to $100, a range that sits entirely beneath the prior quarter's band. Expectations embedded in the price look subdued. That lowers the bar for relief but also signals little patience for another sequential slide. Updates on the Hong Kong dual primary listing and possible Stock Connect inclusion could add support at the margin.
Ultimately, the narrative hinges on one question. Baidu needs to show that its AI cloud acceleration is translating into stabilizing total revenue and recovering margins, not merely impressive growth rates on a smaller base. If cloud strength and an ApolloGo recovery can absorb the planned advertising weakness, the AI transition story regains credibility. If not, investors may keep valuing Baidu on its shrinking legacy business.