Author｜Li Xin

Editor｜Chen Xiaoran

On July 16, Baidu officially announced the initiation of the process to transition to a dual primary listing in Hong Kong from its secondary listing. After the switch is completed within the year, the listing statuses in Hong Kong and the United States will be mutually independent, aiming for inclusion in the Southbound Trading Link adjustment window in September.

Capital Upgrade

The adjustment of the listing structure is a core measure for Baidu to refine its global capital landscape and break free from valuation constraints.

Previously, Baidu adopted a model of "primary listing on NASDAQ + secondary listing in Hong Kong," where the trading status of its Hong Kong shares was dependent on its U.S.-listed entity. This not only failed to meet the regular access requirements of the Hong Kong stock market and hindered access to Southbound funds but also subjected its overall valuation to the sentiment fluctuations of U.S.-listed Chinese concept stocks, preventing the independent value of its new AI businesses from being fairly priced by the market.

According to an announcement disclosed by the Hong Kong Stock Exchange, Baidu's board of directors has fully authorized management to proceed with the conversion, with all compliance procedures expected to be completed by 2026. After completion, Class A ordinary shares and American Depositary Shares will still be freely convertible and tradable across markets.

From a strategic perspective, after the dual primary listing is implemented, Baidu will gain autonomous pricing rights in the Hong Kong stock market. If it successfully passes stringent evaluations such as market capitalization and average daily trading volume, it is expected to follow the path of Bilibili and NetEase and be included in the September adjustment list for the Southbound Trading Link.

The market has already responded. As of the close of the Hong Kong stock market on July 16, Baidu's share price closed at HK$110.3, up 2.6% for the day, with a total market capitalization of HK$300.2 billion.

It should be noted that the current valuation is still anchored to traditional search advertising business. After the entry of Southbound funds, institutions may reevaluate Baidu's growth assets such as AI cloud, large models, and autonomous driving, potentially catalyzing an upward movement in the stock price.

Business Transformation

Behind the capital moves lies a fundamental shift in Baidu's business structure. Relying on its first-quarter 2026 financial report, it is evident that Baidu has completed a historic switch in its business structure, with AI business surpassing advertising business for the first time to become the new main revenue driver.

In the first quarter of 2026, Baidu's general business revenue, excluding iQIYI, was RMB 26 billion, up 2% year-on-year. Among this, AI-driven business revenue was RMB 13.6 billion, accounting for 52%, with a year-on-year increase of 49%.

In contrast, revenue from core traditional search advertising was only RMB 10.2 billion, down 28.6% year-on-year. User data also showed pressure, with Baidu App's monthly active users in March at 655 million, a year-on-year decrease of 69 million, as user traffic and advertising budgets continued to be diverted to platforms like Douyin and Xiaohongshu.

The path to AI monetization is not smooth. As China's earliest company to deeply engage in AI industrialization, Baidu has invested an average of over 15% of its annual revenue in R&D over the past decade, with a cumulative investment of RMB 180 billion in the AI sector. However, it has long been trapped in industry evaluations of "early Layout (layout) but slow implementation."

ERNIE Bot, as China's first general-purpose large model comparable to overseas products, has underperformed in commercialization among the general public on the C-end.

Cross-border businesses such as food delivery and automotive have gradually contracted, and the buzz in the autonomous driving sector has been gradually surpassed by Huawei and Horizon Robotics.

Management once adhered to a closed-source model route, underestimating the rise of open-source models like DeepSeek. Coupled with the departure of core talent from its early Silicon Valley AI lab to start their own ventures, multiple factors have dragged down the pace of commercialization.

After facing these shortcomings head-on, Baidu is accelerating its corrections. Li Yanhong once admitted in an internal meeting that "starting early but arriving late" is an objective reality, compelling the company to shrink (contract) non-core sectors and focus on the full-stack AI mainline of "cloud-model-agent."

Currently, Baidu's core growth mainly comes from AI infrastructure business. In the first quarter, AI cloud revenue was RMB 8.8 billion, up 79% year-on-year, with GPU cloud revenue surging by 184%, securing a large number of computing power procurement orders from government, enterprise, financial, and automotive sectors.

Apollo Go's fully driverless order volume reached 3.2 million, up 120% year-on-year. AI-native marketing, enterprise intelligent agents, and other application scenarios have also steadily landed, officially transitioning AI business from a pure investment phase to a scaling (scaled) monetization phase.

Practical Constraints

Against the backdrop of China's AI industry, while industry demand expansion provides growth opportunities for Baidu, issues such as fierce competition in the sector, imbalanced commercialization structure, and entry of cross-border players pose practical constraints to its breakthrough.

Favorable industry factors are accumulating, with the global large model wave driving a surge in domestic government and enterprise computing power procurement demand. In the first half of 2026, total financing in China's AI sector exceeded RMB 300 billion, with the large model sector alone accounting for about RMB 159.8 billion.

At the same time, the acceleration of industrial digital transformation continues to release orders from central enterprises, financial institutions, and manufacturing companies, with Baidu having a competitive advantage in undertaking these projects due to its years of technological accumulation.

Baidu's differentiated barrier lies in its full-link collaboration capabilities. Relying on the massive real-world scenario data accumulated from its search business, it has long provided training materials for the iteration of the ERNIE large model. Coupled with its AI cloud ranking first in China's AI public cloud market for six consecutive years, Baidu can establish a closed loop of "computing power services-model invocation-industry implementation," capturing market share in the To B government and enterprise services sector.

However, business shortcomings are also prominent. Baidu's AI revenue structure is highly concentrated in the infrastructure layer, with high-margin businesses such as AI applications and enterprise intelligent agents accounting for only 35% of total AI revenue. User penetration of C-end products is insufficient, and profit quality needs improvement.

Industry competition pressure continues to intensify. Huawei and Alibaba Cloud are seizing the government and enterprise cloud market with their ecosystem binding advantages. Large model vendors such as Zhipu AI and DeepSeek are both procuring computing power and developing their own supporting services, gradually cutting into the downstream market to share profits. Overseas, OpenAI, in collaboration with hardware vendors, is launching customized inference solutions, further squeezing the survival space of third-party AI service providers.

In addition, the autonomous driving business is constrained by local road testing regulations, with uncertainty surrounding the pace of scaled implementation.

Continuous large capital expenditures on AI R&D, if coupled with a faster-than-expected decline in traditional advertising business, will squeeze the overall cash flow safety margin.

In summary, Baidu's switch to a dual primary listing is a key strategic move to leverage capital tools to activate its long-term R&D accumulations and reshape market valuation logic.

From its early diversified Layout (layout) to its current focus on contracting its AI main business, Baidu is correcting its past strategic dispersion. However, challenges such as the contraction of traditional main businesses, imbalanced business profitability structure, and intensified industry competition cannot be resolved in the short term.

For management, the only way to completely shed the label of "starting early but arriving late" and gain a firm foothold in the AI industry wave is to rely on the Southbound Trading Link to introduce long-term funds, accelerate the commercialization of AI applications, optimize the revenue structure, and deliver on technical value with continuously implemented operational data.