NextFin News - Baidu is no longer asking investors to treat artificial intelligence as a side project. The company says AI-related revenue has reached 50% of its mix, while its first-quarter 2026 results showed Apollo Go delivered 3.2 million fully driverless rides and had surpassed 22 million cumulative public rides by April. Those figures matter because they show Baidu trying to reprice itself as a full-stack AI company, not just a search-and-advertising business with an AI label attached.
The shift is important for one simple reason: Baidu is now making the case that its value comes from controlling more of the AI chain than most of its peers. Its own disclosures point to a business built around chips, cloud, models, applications, and autonomous driving. The company says revenue from AI applications was RMB 2.5 billion in the first quarter of 2026, while cost of revenue rose to RMB 19.6 billion, reflecting higher spending tied to AI cloud operations. The numbers suggest that Baidu is not merely testing AI products. It is reorganizing the business around them.
That makes the latest comments from finance chief Haijian He more than a branding exercise. They help explain how Baidu wants the market to think about the company’s economics. The message is that AI is becoming the core driver of revenue, and that the advantage lies in integrating the stack rather than relying on a single product or platform. For a company long associated with search, that is a major identity change.
The timing matters as well. China’s internet sector remains under pressure from slower legacy growth, intense competition, and heavier capital requirements for generative AI. In that context, Baidu is trying to show that AI can do two things at once: support top-line growth and build strategic depth. The first quarter offered some proof. Total revenue was RMB 32.1 billion, down 2% quarter over quarter, but the company said its core AI-powered business reached RMB 13.6 billion, up 49% year over year, and for the first time accounted for more than half of Baidu General Business revenue.
That is the core of the story. Baidu is not just arguing that it has AI products. It is arguing that AI now sits at the center of the company’s revenue model. Whether that creates durable leverage or only a more expensive operating structure is the question the market now has to answer.
AI Is Becoming the Center of Baidu’s Business Mix
The most striking fact in Baidu’s latest messaging is the 50% figure. Once AI-related revenue becomes half of the mix, the company is no longer best understood as a legacy internet operator experimenting with new tools. It is an AI-first business that still carries legacy assets. That distinction matters because it changes how investors should think about growth, margins, and capital allocation.
Baidu’s first-quarter results gave that shift some support. The company said core AI-powered revenue reached RMB 13.6 billion, up 49% year over year, and that AI applications generated RMB 2.5 billion in revenue. Apollo Go added another layer of evidence. The autonomous ride-hailing service delivered 3.2 million fully driverless rides in the quarter, with weekly rides peaking at more than 350,000 in March. By April, cumulative rides provided to the public had exceeded 22 million.
Those are not trivial figures. They show that Baidu’s AI portfolio is producing activity at commercial scale, not just demos and research prototypes. They also help explain why management is increasingly comfortable describing the company as a full-stack AI player. The stack is not just theoretical. It spans the chip layer, the cloud layer, the model layer, the application layer, and the mobility layer.
But the 50% milestone should not be confused with proof of profitability. “AI-related revenue” is broad, and the category can include businesses with very different economics. Cloud services, application software, and autonomous driving do not generate the same margins. A larger AI mix is strategically important, but investors still need to know which pieces are growing fastest, which pieces carry the best return on capital, and which pieces may still need heavy investment before they contribute meaningfully to profits.
That is why the top-line shift is more compelling than it is conclusive. It tells us Baidu has crossed a threshold in business composition. It does not yet tell us whether the new mix will ultimately be worth more than the old one.
Haijian He said Baidu’s AI-related revenue has reached 50%, highlighting the scale of the company’s integrated approach across chips, cloud, models, and applications.
The key takeaway from that claim is not simply that AI is growing. It is that Baidu wants the market to value integration itself as a competitive asset. That is a more ambitious argument, and a much harder one to prove.
The Full-Stack Pitch Is Only Valuable If It Creates Leverage
Baidu’s full-stack pitch is plausible because it has real pieces to it. The company has built its own chip capabilities, invested in AI cloud services, developed foundation models, and pushed those assets into consumer and enterprise applications. In theory, that should let Baidu keep more of the economics inside its own ecosystem, reduce dependency on outside suppliers, and iterate faster than rivals that rely on a more fragmented partner network.
In practice, the strategy only works if integration produces measurable leverage. Owning more of the stack is not automatically better. A chip business matters only if the chips are competitive and deployable at scale. A cloud business matters only if it can win customers without sacrificing margins too aggressively. A model business matters only if it can attract traffic, developers, or enterprise demand. An application business matters only if it can convert technical capability into recurring revenue.
Baidu is trying to show that these pieces reinforce one another. That is the right strategic logic. The question is whether the operating results will confirm it. The company disclosed that cost of revenue was RMB 19.6 billion in the first quarter, up 7% quarter over quarter, mainly because of higher AI cloud costs. It also said operating income was RMB 3.2 billion and operating margin was 10%, while non-GAAP operating margin was 12%. Those numbers suggest a company still balancing heavy investment with partial monetization.
That balance is where the market will focus. If AI spending is rising faster than revenue quality, the full-stack story can become a capital-intensive one. If revenue can scale faster than costs, the stack starts to look like a genuine moat. The difference between those two outcomes is what will decide whether Baidu is creating leverage or just complexity.
Baidu’s history makes that distinction especially important. The company has long been technically capable, but investors have often asked whether it could convert those capabilities into durable commercial advantage beyond search and advertising. The full-stack AI argument is an attempt to answer that by showing that the company is not dependent on one product cycle. It is trying to build an operating system for AI services that can span multiple use cases and multiple revenue streams.
That is why the cloud business matters so much. If external customers are paying for Baidu’s AI infrastructure, the model is less circular and easier to evaluate. If the company is mostly internalizing its own workloads, the economics become harder to judge from the outside. The first-quarter disclosures point to real external activity, but they do not yet settle the margin question.
The company’s first-quarter filing said higher AI cloud costs were the main driver of the rise in cost of revenue, showing that the full-stack strategy still requires substantial investment.
That is the trade-off in plain terms. Full-stack AI can create more ways to win, but it also creates more places where the economics can disappoint.
Robotaxis Give Baidu A Real-World Test Of The AI Stack
Autonomous driving is the most tangible part of Baidu’s AI strategy because it turns the stack into a visible service. Apollo Go is not just a software project. It ties together hardware, mapping, perception, decision-making, and fleet operations in a way that gives Baidu a real-world commercialization test. The quarter’s 3.2 million fully driverless rides show that the business is operating at a scale that can no longer be dismissed as experimental.
That does not mean robotaxis are easy money. They are one of the hardest AI-adjacent businesses to scale profitably because they require heavy infrastructure, regulatory patience, technical reliability, and long investment horizons. But they do matter strategically. If Baidu can prove that Apollo Go can expand beyond pilot markets and keep building ride volume, it strengthens the argument that the company’s AI capabilities are not confined to cloud contracts or product demos.
By April, Apollo Go had passed 22 million cumulative rides to the public. Weekly rides also peaked at more than 350,000 in March. Those figures show momentum, and momentum matters in a business that depends on repeated use, public trust, and regulatory progress. The company also said Apollo Go continued to expand internationally, with open-road testing in Switzerland and planned testing in London with Uber and Lyft.
That international ambition matters because it suggests Baidu is trying to turn Apollo Go into a cross-border product rather than a China-only project. If that strategy works, it gives the company another way to monetize its AI stack and another channel to prove technological credibility. If it does not, Apollo Go still remains useful as a demonstration of what Baidu can build with its integrated capabilities.
Still, the key question is not whether Apollo Go is interesting. It is whether it becomes economically meaningful. A high ride count is encouraging, but it is not the same as profit contribution. The market will want to know whether expansion improves unit economics, how much capital the fleet requires, and how quickly the service can scale without eroding returns.
That is why robotaxis should be read as both evidence and test. They show that Baidu’s AI stack extends into a physical business with real operations. They also show how far the company still has to go before the stack becomes a self-sustaining profit engine.
Baidu said Apollo Go delivered 3.2 million fully driverless rides in the first quarter and had surpassed 22 million cumulative public rides by April, underscoring the scale of its autonomous-driving push.
The strategic value is clear. The commercial value still has to be proven.
Why Baidu’s Repricing Story Matters Beyond One Company
Baidu’s full-stack AI pitch matters beyond its own earnings because it is helping define how the market will value AI businesses in the next phase of the cycle. Investors are increasingly asking whether the most valuable AI companies will be those that own models, those that own cloud capacity, those that own applications, or those that control all three. Baidu is trying to argue that the answer is all of the above.
That is a powerful narrative, but narratives only last if the numbers keep supporting them. The company’s first-quarter results showed a business that is still under pressure in its legacy areas while AI-linked lines grow enough to offset some of that weakness. Total revenue was RMB 32.1 billion, and general business revenue was RMB 26.0 billion. The AI side is now large enough to matter to the whole company, but it is still being built inside an environment that remains capital intensive and highly competitive.
So the market’s job is not to choose between “Baidu as search company” and “Baidu as AI company.” It is to decide how much of the latter is real, how much is scalable, and how much deserves a premium. That is a more complicated valuation exercise than simply attaching an AI label to the stock.
The next few quarters should bring more evidence. Investors will watch the pace of AI revenue growth, the mix within that revenue, the margin trajectory of AI cloud, and whether Apollo Go keeps expanding ride volume and geography. They will also look for signs that the company can keep its AI push from overwhelming the economics of the rest of the business.
For now, Baidu has made its case clearly: it believes the stack itself is the strategy. The market still has to decide whether that stack turns into leverage or just another expensive way to stay in the race.
The bottom line is simple. Baidu has already moved beyond the question of whether it is an AI company. The harder question is whether being a full-stack AI company will be more profitable than being a narrower one.
Explore more exclusive insights at nextfin.ai.
