Shenwan Hongyuan released a research report upgrading Baidu Group-SW (09888) to a "Buy" rating. The firm projects Baidu Group's total revenue for 2025-2027 to be RMB 128.5 billion, RMB 133.1 billion, and RMB 141.0 billion, respectively. Under the SOTP valuation method, the group's total target valuation is RMB 430.2 billion, corresponding to a target price of HK$172.54 per share for its Hong Kong-listed stock. Shenwan Hongyuan's key viewpoints are as follows.
In the AI era, domestic AI cloud revenue is accelerating. Starting from Q3 2024, leading cloud providers like Alibaba and Tencent have increased their AI capital expenditures. In the first three quarters of 2025, the capital expenditure/revenue ratio for BAT surpassed 10%, approaching the level of overseas cloud providers in 2023. Since early 2025, open-source, low-cost large models have driven demand for AI cloud services, leading to a recovery in growth for providers like Alibaba Cloud. A horizontal comparison shows that internet cloud providers each have distinct advantages in their full-stack layouts: Tencent and ByteDance excel in AI applications, while Alibaba and Baidu control the entire technology stack from chips to industry applications and can provide end-to-end services, with Google Cloud serving as the benchmark for full-stack AI overseas.
Baidu continues to advance its full-stack AI development, with its Intelligent Cloud business experiencing accelerated growth. At the 2025 Baidu World Conference, Baidu launched the Kunlun Core M-series chips, the Omni-modal ERNIE 5.0 model, and multiple enterprise-grade intelligent agent applications. In Q3 2025, Baidu's AI cloud revenue reached RMB 6.2 billion, with AI infrastructure revenue increasing 33% year-over-year and accelerator subscription revenue surging 128%. For the first three quarters of 2025, Intelligent Cloud revenue totaled RMB 19.33 billion, a 31% year-over-year increase, with cloud revenue accounting for 37% of Baidu Core revenue. At the industry level, Baidu Intelligent Cloud held the top market share in AI large model solutions in H1 2025 and ranked among the top three in both GenAI IaaS and the large model public cloud market.
The self-developed AI chip, Kunlun Core, is expected to enter a volume expansion phase. Having progressed through stages of technological foundation, large-scale implementation, and large model adaptation starting in 2024, it has now formed an integrated training-inference product matrix from the cloud to data centers. In 2025, the Tianchi 32 and 64 super-nodes were launched, with the Tianchi 512 super-node—capable of supporting 512-card interconnection and trillion-parameter training—scheduled for deployment in 2026. The M100 chip for large-scale inference will also be introduced, and the M300 chip, aimed at addressing shortcomings in domestic high-end training, is planned for launch in 2027, with a target of achieving a million-card single Nvidia GPU cluster by 2030. Commercially, Kunlun Core has already penetrated sectors including internet, finance, and energy, entering the systems of major clients like China Merchants Bank and China Southern Power Grid, and it won a nearly RMB 1 billion server procurement order from China Mobile.
In intelligent driving, Apollo Go's order volume is surging, with profitability significantly improving. After more than a decade of dedicated effort, Apollo Go possesses a complete full-stack technology system encompassing vehicles, roads, cloud, and maps. Benefiting from enhanced economies of scale and substantially reduced costs, per-vehicle profitability has turned positive. Commercialization accelerated significantly in 2025, with weekly orders exceeding 250,000, leading globally. Q3 2025 orders reached 3.1 million, a more than 200% year-over-year increase, with expansion into overseas regions like Europe and the Middle East, which feature higher average order values. The firm believes domestic Robotaxi holds significant advantages over overseas counterparts in terms of cost and infrastructure and is optimistic about Baidu's strategic layout for the next-generation "mobile living space."
Risk warnings include intensified competition in the advertising industry, macroeconomic changes affecting corporate demand for cloud services, heightened competition among domestic cloud providers, uncertainties in innovative businesses, and regulatory risks.
Comments