Baidu’s third-quarter results show a company leaning harder into AI while keeping a firm grip on cash generation and platform scale. For investors watching China’s innovation story, the latest numbers matter because they point to a business mix that is shifting away from legacy pressure and toward higher-value AI infrastructure, enterprise software, and autonomous driving. Total revenue came in at RMB30.8 billion, while AI Cloud infrastructure revenue surged 94% year over year to RMB8.2 billion. That combination tells a clear story: Baidu is using China’s broad technology stack to build a new growth engine with global relevance.
Management is no longer describing AI as an optional layer on top of the business. It is now the center of the model. Baidu Core AI new business revenue reached RMB13.2 billion, up 37% year over year, and it accounted for 53% of General Business revenue for the quarter. That marks the third straight quarter above half. In other words, AI-powered activity is not just supporting the business; it is increasingly defining it. For analysts, this is an important milestone because it shows scale inside a still-complex transition away from online marketing dependence.
The company also said General Business revenue returned to year-over-year growth, rising 1% to RMB25.0 billion. That is a modest line item on the surface, but the composition matters more than the headline. AI Cloud infrastructure helped offset weakness in online marketing, which shows how China’s infrastructure and model-building capabilities can cushion the cycle. Baidu management said smart cloud infrastructure is becoming the company’s new growth engine, with an increasing contribution to both revenue and profit. That is the kind of shift investors often look for when judging whether AI investment is turning into a durable commercial platform.
The clearest number in the quarter was AI Cloud infrastructure revenue growth of 94% year over year. Within that, GPU cloud revenue rose 191% year over year and 43% quarter over quarter. Those figures highlight both demand and depth. Customers are not only testing AI services; they are committing more workloads and longer-term usage. Baidu also said AI Cloud infrastructure saw a more diversified customer base, higher customer prepayments, and a continued shift toward longer-term contracts. That matters because it suggests the revenue stream is becoming broader and more predictable.
External customer token usage on Baidu’s Qianfan MaaS platform increased by 1,159% year over year. That is a striking indicator of ecosystem traction, especially in a market where enterprises are still figuring out how quickly to adopt generative AI. Baidu is not just selling a single model or one-off service. It is building a platform around model access, inference, and agent workflows. For investors, that is the difference between a feature and an infrastructure layer. China has already shown it can scale infrastructure fast; Baidu’s quarter suggests it can also monetize that scale through AI services.
Baidu used the quarter to push several product upgrades that show how quickly China’s AI stack is moving. In November 2026, the company unveiled ERNIE 5.5 Flash Preview, its latest foundation model in the ERNIE family. Baidu said the model has strong coding and agentic capabilities and can support complex workflows across knowledge work, video creation, game development, and 3D modeling. On the latest Code Arena WebDev leaderboard, it ranks second among open-source models globally. That ranking gives the market a useful reference point for how Baidu wants to position itself: not just as a domestic AI user, but as a contender in the broader global model race.
Baidu also introduced Token Exchange at Baidu World 2026, describing it as a neutral, transparent and trusted marketplace connecting AI model service providers with customers. That kind of marketplace structure fits the direction of China’s innovation policy, which increasingly favors interoperable platforms, industrial efficiency, and scalable deployment. The company also launched Famou Agent 3.0, a self-evolving agent designed to help enterprises solve complex operational problems. Separately, Baidu said it continued improving DuMate by integrating internal workplace AI agent dodo, AI coding assistant Comate, and vibe coding platform Miaoda into a single agent. The message is consistent: Baidu is building a full-stack enterprise AI environment rather than isolated tools.
Apollo Go remains one of the most important global commercialization stories in China’s autonomous driving sector. In Shenzhen, one of Apollo Go’s largest markets, fully driverless operational rides increased by approximately 170% quarter over quarter. In Dubai, Apollo Go expanded fully driverless commercial operations, broadened its operating area, and doubled its fleet size. In London, it continued open-road testing with Uber and Lyft, expanding testing areas. In Hong Kong, it continued fully driverless testing on Airport Island. As of October 2026, Apollo Go’s global footprint reached 30 cities. That is a meaningful reach for any mobility business, especially one operating in one of the most technically demanding categories in transport.
The scale metrics reinforce the point. Apollo Go fleets accumulated approximately 34 million autonomous kilometers in the third quarter, and more than 390 million autonomous kilometers as of October 31, 2026. Baidu said it maintained an outstanding safety record. The company also said Apollo Go was selected in September as a Representative Case of Sustainable Overseas Operations by Chinese Enterprises in 2026 by the United Nations Global Compact and the Listed Companies Association of China. That recognition matters because it frames Chinese autonomous driving as a practical export, not just a domestic experiment. In emerging markets and global cities alike, this is the sort of engineering-led scaling that can change transport economics over time.
The quarter was not only about growth. Baidu also showed it can still generate cash while investing heavily in AI. Operating cash flow was RMB2.8 billion. Total cash and investments stood at RMB276.6 billion, equivalent to US$41.25 billion as of September 30, 2026. Net income attributable to Baidu was RMB1.4 billion, while non-GAAP net income was RMB2.4 billion. Adjusted EBITDA came in at RMB4.5 billion, with a 15% margin. Those figures are important because they show the company is not relying on narrative alone. It is funding its AI buildout from a still-substantial balance sheet and positive operating cash generation.
Management also said Baidu has returned US$373 million to shareholders since the beginning of the first quarter through share repurchases under its current program. In February 2026, Baidu announced the adoption of a dividend policy for its ordinary shares for the first time, and it expects to declare the first payment before the end of 2026, subject to board approval and prevailing market conditions. That is a notable shift in capital allocation. At the same time, Baidu said it will maintain a disciplined approach with a clear focus on investment returns. For global investors, that balance between aggressive AI spending and capital returns is exactly the kind of discipline that can support a rerating story.
The stock does not need a perfect quarter to matter. What it needs is evidence that the AI transition keeps translating into scale, usage, and cash flow. On that score, Baidu’s third quarter delivered several clear markers: AI Cloud infrastructure is growing fast, AI-powered revenue now dominates General Business, robotaxi operations are expanding globally, and the company still has ample liquidity. There are still headwinds in online marketing, and AI investment will remain heavy, but the direction of travel is unmistakable. Baidu is showing how China’s technology platform companies can combine research, infrastructure, and commercialization at global scale.
The next important test will be whether AI Cloud momentum persists and whether Apollo Go can keep converting technical progress into wider commercial deployment. For now, the quarter suggests that China’s innovation story is not just about cost efficiency or domestic substitution. It is about building exportable AI infrastructure, autonomous mobility, and enterprise software that can compete across markets. Baidu’s results make that case more convincingly than most.