Baidu has really made AI applications a success.

At the beginning of this year, DeepSeek emerged suddenly, leading foreign institutional investors to shift their focus collectively and turn bullish on Chinese assets. Among them, technology leading stocks such as Baidu Group, SMIC, Meituan, and BYD, which are often crowned as the "Top Ten Stocks in Hong Kong" by the market, are the representative targets.

However, the previous "endorsement" from foreign investors remained at the research level. To re - enter the Chinese stock market, fundamental support is still needed. Therefore, the first - quarter earnings reports of various companies, especially technology companies this year, have become the key variables that the capital market focuses on. Just recently, the release of Baidu Group's first - quarter earnings report may have boosted the sentiment of the hesitant funds.

The first - quarter report shows that during the reporting period, Baidu's total revenue reached 32.5 billion yuan, and the net profit of Baidu Core increased by 48% year - on - year to 7.63 billion yuan.

Meanwhile, Baidu's cloud business performed impressively, with a growth rate of 42%. In addition, Apollo Go has entered Dubai and Abu Dhabi successively. As of May, it has provided over 11 million travel services globally.

This shows that the dual - engine effect of "cloud + autonomous driving" set by Baidu in AI applications has emerged.

01

The cloud business continues to strengthen, fulfilling the possibility of "exceeding expectations"

Affected by the "DeepSeek market" at the beginning of the year, technology stocks experienced an explosive rise. Now that the "bonus period" has passed, the Hong Kong stock market, which hosts most technology companies, has actually shown differentiation. Whether they can continue to rise in the future depends on whether the company's fundamentals can meet or exceed expectations.

Looking at the fundamentals, the core lies in the performance. Based on the first - quarter performance, Baidu's cloud business has been growing for several consecutive quarters.

With the integration of large models into domestic ecosystems such as social APPs and the increase in AI interaction scenarios, the demand for computing power has skyrocketed. The market unanimously predicts that the demand for AI - related cloud services is expected to rise.

The development of Baidu's cloud business is a confirmation of the market's expectation that "AI demand drives cloud growth".

Achieving the market's performance expectations relies on technological drive. To gain the recognition of users and developers, it's all about having stronger capabilities and lower costs.

From March to April this year, Baidu successively released four large models. Among them, ERNIE Bot 4.5 Turbo demonstrated excellent multi - modal capabilities, with an average score higher than that of GPT - 4o, and a lower price. The input price per million tokens is only 0.8 yuan, and the output price is 3.2 yuan. ERNIE X1 Turbo is the first in - depth thinking model that independently uses tools, with a price only 25% of DeepSeek - R1. The input price per million tokens is 1 yuan, and the output price is 4 yuan, offering higher cost - effectiveness.

Meanwhile, Baidu Smart Cloud Qianfan Large Model Platform has connected with hundreds of mainstream large models, with lower calling prices. It announced compatibility with the MCP protocol to help developers and enterprises flexibly select tools and efficiently connect to intelligent agents.

On the other hand, at the framework level, PaddlePaddle 3.0 has achieved multiple technological breakthroughs such as automatic parallelism, neural network compilers, and high - order automatic differentiation, significantly optimizing the training efficiency of large models. In addition, Baidu lit up the third - generation Kunlun Chip cluster of ten thousand cards, ensuring the stability and energy efficiency of training tasks.

The end - to - end optimization helps Baidu further expand its user base and application scenarios. The market's recognition of Baidu's full - stack artificial intelligence capabilities is reflected in project tenders.

Market data shows that in the large - model tendering and bidding market in Q1 2025, Baidu Smart Cloud led the industry with 19 winning projects and a winning bid amount of 450 million yuan. Both core indicators ranked first among general large - model manufacturers. The cooperation partners are concentrated in leading enterprises such as China Merchants Group and "Tianguang Robot".

02

Leading in intelligent driving, going global to consolidate the first - mover advantage

In addition to the continuous strengthening of the cloud business, autonomous driving, one of the "dual engines" of Baidu's AI applications, also continued its outstanding performance in this quarter.

The financial report data shows that in the first quarter of this year, Apollo Go provided over 1.4 million travel services globally, a year - on - year increase of 75%. As of May, Apollo Go has provided over 11 million travel services globally.

Autonomous driving is one of the most promising AI application scenarios. In this year's investment themes centered around AI applications, autonomous driving is also one of the key tracks. Institutional researchers believe that 2025 may be the year when automotive intelligence crosses the critical point, and AI technological breakthroughs and "equal rights for intelligent driving" are expected to drive the rapid growth of industry penetration.

Against this background, the layout and implementation of Baidu's autonomous driving travel service platform, Apollo Go, are timely.

First of all, in 2024, Apollo Go was the first to release the world's first L4 - level autonomous driving large model, Apollo ADFM. So far, the cumulative safe test mileage has exceeded 130 million kilometers, and the accident rate is only one - fourteenth of that of human drivers, leading the entire industry in terms of safety, with a safety level close to that of the domestic large aircraft C919.

Cathie Wood, the founder of the globally renowned investment institution ARK Invest, also said not long ago that "the world's top players in autonomous driving are Waymo, Tesla, and Baidu in the United States." Meanwhile, the funds under ARK Invest further increased their holdings of Baidu stocks.

Secondly, the sixth - generation driverless vehicle of Apollo Go, which was launched simultaneously with Apollo ADFM, is equipped with Baidu's sixth - generation intelligent system solution. The mass - production cost has dropped to 204,600 yuan per vehicle, only one - seventh of Waymo's operating cost. The significant reduction in single - vehicle cost means that Apollo Go has strong large - scale deployment capabilities. Solving the problems of cost and scale brings it closer to a clear profit model.

Looking at the international market, Apollo Go is also exporting the data and experience accumulated in domestic cities overseas.

In March this year, Apollo Go successively announced the launch of large - scale autonomous driving tests and services in Dubai and Abu Dhabi. According to market news, Switzerland and Turkey will follow suit.

When entering Hong Kong before, Apollo Go was the first company approved to test autonomous driving on open roads. In April this year, its test area in Hong Kong was further expanded. Now, accelerating its entry into the Middle East market, Apollo Go is expected to consolidate this first - mover advantage.

Against the background of domestic supply chains going global collectively, overseas logistics and other infrastructure are becoming increasingly perfect, and governments of many countries have successively given the green light to autonomous driving. Apollo Go's overseas journey can be regarded as going with the trend.

Recently, Apollo reached a strategic cooperation with Shenzhou to jointly create the world's first autonomous driving car rental service. This move seems to foreshadow more possibilities for the commercialization of Baidu's autonomous driving business.

03

In the first year of AI applications, Baidu takes the lead in implementation

Cathie Wood, who firmly increased her holdings of Baidu, once predicted that "2025 will be a crucial year for autonomous driving." Similarly, multiple research reports have also proposed the view that "2025 is expected to be the first year of AI applications." Institutional investors have also included more AI - application - related targets in their investment strategies.

Both product and capital predictions are looking forward to an explosion of AI applications this year.

"Developing towards applications" has long been the direction of Baidu's AI strategy.

In fact, through the open - sourcing and iteration of ERNIE Bot and the cloud platform, it can be seen that Baidu's starting point is to solve the needs at the application layer.

For example, the newly released ERNIE Bot 4.5 Turbo and ERNIE Bot X1 Turbo not only have multi - modal and strong reasoning characteristics but also emphasize cost - effectiveness. At the press conference, Robin Li, the founder and chairman of Baidu, said that after the cost is reduced, developers and entrepreneurs can develop with confidence, and enterprises can deploy large models at low cost, ultimately promoting the explosion of applications in various industries.

In the first quarter of this year, Baidu Smart Cloud Qianfan Large Model Platform was upgraded again, adding an expanded model library and a more comprehensive toolchain to support the training and fine - tuning of multi - modal models and in - depth thinking models, further promoting the development of AI applications.

The 42% single - quarter growth rate of the cloud business proves the correctness of Baidu's "developing towards applications" path.

Meanwhile, the test results of Apollo Go in multiple domestic cities and overseas regions, as well as the performance of Baidu's autonomous driving business, are also a strong footnote to Baidu's successful implementation of large - model capabilities in AI applications.

The continuous strengthening of the cloud business and Apollo Go's acceleration into overseas markets jointly support Baidu's second and third growth curves. And Baidu's AI capabilities honed over the years are also reshaping its other applications.

For example, Baidu Search, as the cornerstone business, continued to focus on improving the quality of AI content in the first quarter of this year. By integrating the full - version of DeepSeek - R1 and the online search function, it coped with the challenges of high concurrency and high bandwidth. In March 2025, the MAU of Baidu APP increased by 7% year - on - year to 724 million.

In addition, as of the end of the first quarter, the MAU of Baidu Wenku's AI function reached 97 million, making it one of the AI products with the most paying users globally. The MAU of Baidu Netdisk's AI exceeded 80 million, indicating initial success in its AI transformation.

At the developer conference in April, Robin Li released several AI applications in one go, such as high - persuasive digital humans, the general super - intelligent agent Xinxiang APP, and the content operating system Cangzhou OS. The no - code programming tool Miaoda, which was released in November last year, was also opened to the whole society in March this year. Anyone can generate an application with just one sentence through Miaoda.

AI technology improves the user experience, directly increasing the added value of products and user stickiness. As the AI functions of the above - mentioned applications are enhanced, user growth and the user payment rate will also increase, which is expected to drive Baidu's revenue growth.

The continuous growth of Baidu's cloud business in recent years has indirectly reflected the explosive power of AI applications. In the face of the view that this is the first year of AI application implementation, institutional forecasts suggest that AI application revenue is expected to be reflected in the performance of various companies starting from the fourth quarter of this year.

Capital has made early arrangements and increased its holdings of leading technology stocks. Under the argument of valuation repair, foreign capital is prudently flowing back, and is taking the lead in investing in asset targets with relatively more certain returns.

For example, in the first quarter of this year, Bridgewater Associates increased its holdings of Baidu stocks by more than ten times, and Fidelity Investments also increased its holdings of Baidu stocks by nearly ten times.

Since 2025, the market generally believes that the valuation repair of Chinese assets is the core logic of the equity market, and AI demand is the main driving force for the entire investment. Targets related to AI computing power and large models, as well as enterprises related to AI applications, frequently appear in institutional investment strategies, and the overseas expansion of high - quality enterprises is considered a long - term high - growth track.

Under this logic, Baidu almost hits the bull's - eye.