Baidu 2025 Financial Report: AI Transformation Shows Promising Results, Commercialization Inflection Point Accelerates

On February 26, Baidu released its 2025 fourth-quarter and full-year financial results, showing that in 2025 total revenue reached 129.1 billion yuan, and AI business revenue reached 40 billion yuan; in the fourth quarter, Baidu’s total revenue was 32.7 billion yuan, and AI business income accounted for 43% of Baidu’s general business revenue, exceeding market expectations. This set of financial results clearly shows that Baidu, which started as a search engine company, is accelerating its transition to an AI-driven enterprise. AI business revenue has moved steadily from the early stage of technical investment into a key period for large-scale monetization, becoming the core force supporting the company’s sustainable growth.

Large-Scale Monetization of AI Business Takes Hold

Judging from core financial data, Baidu’s full-year 2025 total revenue reached 129.1 billion yuan, a slight year-over-year decrease of 3%. This fluctuation was mainly due to slower growth in traditional businesses such as search advertising. Although traditional businesses face certain pressure, the explosive growth of Baidu’s AI-related business is the biggest highlight of this earnings report. The data shows that in full-year 2025, Baidu’s AI-related revenue exceeded 40 billion yuan, up 48% year over year. Of this, the share of revenue from core AI new businesses in the fourth quarter was as high as 43%. This means Baidu’s revenue structure has successfully shifted from being dominated by traditional advertisers to a new pattern driven by both AI business and traditional business, with a clear and significant transformation effect.

Baidu’s large-scale monetization of its AI business has achieved breakthroughs across multiple sub-sectors. Among them, AI-native marketing services are especially outstanding. Full-year revenue from this segment grew 301% year over year, becoming a new growth engine for Baidu’s advertising business. It is understood that this service leverages the multimodal technical capabilities of the Ernie Bot large model, enabling intelligent generation of marketing content and precise delivery. It has fundamentally changed the operating model of traditional search ads, providing advertisers with more efficient marketing solutions, and fully demonstrates the value of AI technology in upgrading and transforming traditional marketing.

At the earnings call, in response to analysts’ questions about the competitive landscape of large models, Robin Li gave a clear reply: “In competition, we always believe that applications are more important than the model, because the value of the model ultimately needs to be realized through applications.” Robin Li said that recently Baidu released the 2025 upgraded version of the Ernie Bot large model, and it also proactively adjusted its organizational structure by splitting the large-model research and development team and arranging it across different directions: one team continues to tackle frontier capabilities of foundational large models; the other team is more closely aligned with specific business needs and application scenarios, focusing on cost reduction, efficiency improvement, and speed-up. “The ultimate goal is to achieve the best application outcomes.” Robin Li said.

Baidu’s intelligent cloud infrastructure business has also maintained strong growth momentum. In 2025, Baidu’s intelligent cloud generated full-year revenue of 20 billion yuan, up 34% year over year, including subscription revenue from AI high-performance computing facilities, which grew 143% year over year. Behind this growth is the opportunity in the compute market brought by the explosive demand for global AI large-model training, and Baidu, through its long-term layout in AI infrastructure, has successfully captured industry dividends and further strengthened its competitive advantage in the intelligent cloud market. With the continued iteration of AI large models, enterprises’ demand for high-end computing capacity is expected to keep rising, and Baidu’s intelligent cloud is expected to continue benefiting from this industry trend.

At the same time, the commercialization rollout of Baidu’s autonomous driving business (RoboTaxi) has become a benchmark for the application of Baidu’s AI technology. The financial report shows that in the fourth quarter of 2025, RoboTaxi orders reached 3.4 million rides, up 200% year over year. Cumulative service times surpassed 20 million rides, and the business covered 26 cities worldwide. As of now, RoboTaxi’s fully driverless operating mileage has exceeded 190 million kilometers, and it has maintained a good safety record, which the industry has praised as a global example of autonomous driving commercialization. Worth noting is that RoboTaxi is accelerating its global expansion, having already entered overseas markets such as London, Abu Dhabi, and Seoul. It has also reached cooperation agreements with international mobility platforms such as Uber and Lyft, becoming an important calling card for China’s AI technology to go global.

Baidu Enters a New Stage of Value Return to Shareholders

In early February, Baidu announced on the Hong Kong Stock Exchange: the company stated that it would use 5 billion US dollars, about 10% of its current market value, to repurchase shares, and expects to begin paying its first dividend in 2026. This move means Baidu has gradually entered a new stage focused on returning value to shareholders, transitioning from the earlier period of high-investment transformation and breakthrough. As explained by research institutions, this move not only helps stabilize expectations in the capital markets, but can further enhance investors’ long-term confidence in Baidu’s AI transition.

Technology research and development is the core support for Baidu’s transformation. In 2025, the company continued to increase its R&D investment and steadily strengthen its core technology barriers. That year, Ernie Bot 5.0 was officially launched, with 2.4 trillion parameters. It remains industry-leading in the multimodal large model domain, providing a solid technical foundation for top-layer businesses such as AI-native applications, intelligent cloud, and autonomous driving. At the same time, Baidu has made important progress in AI chip development. In January 2026, Kunlun Chip under Baidu officially started the process for an IPO on the Hong Kong stock market. This move is regarded by industry analysts as a key step for Baidu to solve the valuation difficulties in the hardware business. Through independent financing, it will further accelerate AI chip R&D and capacity expansion, and improve Baidu’s full AI industrial-chain layout.

Baidu’s impressive performance in its 2025 financial report may well confirm the arrival of a commercial inflection point for China’s AI industry. Prior to this, most domestic AI companies were still in the stage of technical R&D and concept validation, whereas Baidu successfully achieved large-scale monetization of its AI business through a mature model of “technological breakthroughs + scenario deployment + capital operations.” This provides a transformation path that domestic technology companies can learn from. Baidu’s practice fully shows that the value of AI technology ultimately needs to be realized through industrial scenario deployment. Only by deeply integrating core technologies with real application scenarios such as marketing, cloud services, and autonomous driving can the transformation truly be realized from technological value to commercial value.