Key Points:

1. The battle for cloud computing has extended to airports. At major hubs like Beijing Capital Airport and Shanghai Hongqiao Airport, AI cloud vendors such as Alibaba Cloud and Baidu Intelligent Cloud have dominated the core advertising spaces. Alibaba Cloud aims to capture the market share of AI cloud, while Baidu Intelligent Cloud emphasizes AI implementation across industries.

2. Cloud vendors who invest more decisively in AI will grow faster. This trend is consistently seen in the enterprise sector. The earlier a company embraces AI and completes capability migration, the more likely it is to achieve structural growth in the new technology cycle.

3. When choosing an AI cloud partner, enterprises need to consider whether potential partners have experience in large-scale, multi-industry, and multi-scenario implementations, and whether they can deliver stable and reliable AI capabilities in complex business systems.

4. The key to the true implementation of AI lies in building a full-stack AI cloud infrastructure. This infrastructure is becoming the most important foundational support for enterprises moving towards intelligence.

5. As leading players in the domestic AI cloud market, Baidu Intelligent Cloud and Alibaba Cloud have keenly captured this trend. Baidu Intelligent Cloud's airport advertising showcases its status as a top-tier infrastructure builder, with deep technical expertise, extensive project implementation capabilities, and a full-stack AI cloud infrastructure, making it the choice for more enterprises in their intelligent transformation.

Author: Chang Yuan

Editor: Key Point Master

During a recent business trip, Key Point Master noticed that the battle for AI cloud has extended to airports. As an important offline touchpoint for To B vendors, advertising in the AI cloud sector has significantly heated up at airports. At Beijing Capital Airport, Shanghai Hongqiao Airport, and other major hubs, vendors such as Alibaba Cloud, Baidu Intelligent Cloud, Huawei Cloud, JD Cloud, and China Mobile Cloud have dominated the core advertising spaces. Some focus on large models, some promote intelligent agents, and others showcase industry implementation cooperation cases.

The rise of AI and large models is reigniting growth expectations across various industries, and cloud vendors are taking advantage of this trend to increase advertising investments, compete for users, and capture market share. The greatest value of airport advertising lies in its high-frequency reach to business travelers, especially government and enterprise customers with purchasing power. During this period, airports have almost become exhibition halls for AI cloud.

The advertising copy on different vendors' billboards also reveals their strategic orientations:

Alibaba Cloud: "For AI, use Alibaba Cloud; Tongyi Large Model | Open Source and Globally Leading"

Baidu Intelligent Cloud: "For AI implementation, use Baidu Intelligent Cloud; Leading China's AI public cloud service market share for 6 consecutive years"

It is not difficult to see that Alibaba Cloud hopes to capture the market share of AI cloud, while Baidu Intelligent Cloud aims to defend its position and emphasizes deep AI implementation. With different paths and strategies, the war for AI cloud is becoming increasingly fierce.

The earlier enterprises embrace AI, the faster their revenue growth.

Over the past decade, artificial intelligence in China has evolved through two stages: from "discriminative AI" focused on recognition and classification to "generative AI" characterized by generation and interaction. The early "AI Four Dragons" rose to prominence with speech recognition, image processing, and other technologies, representing a generation of AI technology companies. However, AI at that stage was more like a toolbox with segregated functions, making it difficult to systematically serve enterprise needs.

A turning point emerged after the release of ChatGPT. For the first time, large language models gained the ability to understand human intent and demonstrated leapfrog progress in semantic understanding, reasoning, logic, memory, and other dimensions. AI began to move from handling rule-based problems to participating in dialogues and even executing tasks.

With the accelerated development of domestic large models, models with strong reasoning capabilities and low usage costs, such as DeepSeek, have risen rapidly, continuously raising the upper limit of user experience. By 2025, AI has progressed from the early "usable" stage to the "user-friendly" stage: from writing copy and generating images to creating videos, PPTs, automatic programming, intelligent customer service, and more. An increasing number of enterprises are integrating AI into their business processes, driving organizational restructuring and efficiency improvements.

A key point that is easily overlooked is that AI implementation needs to be built on cloud infrastructure. Cloud infrastructure is to AI what the body is to the brain – the capability boundaries of AI are limited by the computing power, storage, networking, and other resource allocations it relies on. In other words, AI runs on the cloud, and the cloud determines how fast, stable, and far it can run.

Therefore, after the release of ChatGPT, domestic cloud vendors successively launched models, ushering in the model era for the cloud computing industry. Baidu Intelligent Cloud and Alibaba Cloud were the first two players to fully bet on AI. Baidu took the lead in launching the ERNIE series of large models and the ERNIE Bot, a large model product comparable to ChatGPT; Alibaba followed suit with Tongyi Qianwen and chose open source and openness as its differentiation strategy.

We observe that in the cloud computing industry, those who invest more decisively in AI will grow faster. IDC data shows that in the second half of 2024, China's public cloud market grew by 17.7% year-on-year, setting a new two-year high. In the first quarter of this year, Alibaba Cloud's revenue increased by 17.7% year-on-year; Baidu Intelligent Cloud led the pack with a year-on-year growth rate of 42%, becoming one of the few domestic cloud vendors achieving positive momentum.

This trend is also consistently seen in the enterprise sector. The earlier a company embraces AI and completes capability migration, the more likely it is to achieve structural growth in the new technology cycle. For example, banks are reconstructing customer service and risk control processes through large models, manufacturing enterprises are introducing intelligent quality inspection and AI agents to assist in R&D, and retail and e-commerce are using AI to optimize supply chains and marketing rhythms. AI has become an underlying variable determining core business outcomes such as enterprise efficiency, revenue, and profit growth.

If the question of the past decade was "whether to transform", today's question is "how to transform faster".

This is a pressing question. However, at the specific execution level, most enterprises are still in the early stages of exploration. Business owners first need to figure out: faced with the complex AI cloud market, what kind of AI cloud partner should they choose to realize intelligent transformation?

Implementation is the only criterion for testing AI cloud.

The wave of large models has lasted for nearly two years, yet no blockbuster C-end applications have emerged, nor has the APP boom of the mobile internet era been replicated. The reason is not difficult to understand: the true value of large models lies not in C-end consumer scenarios but in specific business scenarios across various industries.

AI is reconstructing the underlying capabilities of enterprises. In the process of intelligent transformation, enterprises usually need to consider three core criteria when choosing an AI cloud partner:

First, absolute stability and reliability. For enterprises, business continuity is a bottom-line requirement. Even a half-hour downtime of a core system may mean production line stoppage, transaction interruption, or even data loss, with dire consequences.

Second, a deep understanding of the business. AI cannot be "universally implemented" and needs to be embedded in vertical scenarios such as finance, manufacturing, healthcare, and government affairs. Each industry has unique know-how, and AI cloud vendors must deeply understand enterprise businesses, providing in-depth solutions tailored to business processes like consultants.

Third, delivering solutions. Essentially, what enterprises purchase is not the technology itself but the ability to use AI to solve practical problems. For example, in the power industry, what enterprises truly need are solutions to problems such as grid security, load scheduling, and power supply configuration, rather than simple model invocations.

What kind of cloud vendor can meet the above criteria? While technology and price are important, engineering capabilities are even more crucial. Enterprises need to focus on whether the chosen AI cloud vendor has experience in large-scale, multi-industry, and multi-scenario implementations and whether it can deliver stable and reliable AI capabilities in complex business systems.

Take Baidu Intelligent Cloud as an example. Its latest brand advertisement states: "Focused on AI cloud for a decade; Kunlun Chip + Baige GPU Cloud + ERNIE Family + Qianfan Large Model Platform = Full-stack AI cloud infrastructure for the intelligent era; Leading in both the number and amount of large model bids won in the Chinese market".

From past disclosures:

First, Baidu is one of the earliest domestic cloud vendors to propose the concept of AI cloud, possessing comprehensive AI technology capabilities from chips to models.

Second, Baidu Intelligent Cloud was also the first to advocate "cloud-intelligence integration", emphasizing the deep integration of AI capabilities with industries to drive implementation. Judging from data over the past few years, it has firmly maintained its position as the leader in China's AI public cloud service market share.

Third, Baidu Intelligent Cloud was the first to launch Qianfan, a domestic large model platform, and has been emphasizing full-stack AI infrastructure since then. Public data shows that it leads in both the number and amount of large model bids won in China.

Baidu Intelligent Cloud's engineering implementation capabilities are undoubted, having been verified on a large scale across multiple industries. From frequent public information, Baidu Intelligent Cloud serves 65% of central enterprise customers, TOP 15 new energy vehicle enterprises, and more than half of game manufacturers, and widely supports AI innovation in government affairs, manufacturing, energy, and other industries.

Alibaba Cloud's advertisement focuses more on its AI cloud positioning and leading open-source models. The open-source strategy has built a vast developer ecosystem for Alibaba Cloud, which is now constructing a collaborative system of "infrastructure-AI platform-commercial implementation" with computing and networking as the foundation, open-source models as the engine, and scenario innovation as the breakthrough.

The key to implementation lies in building full-stack AI cloud infrastructure.

Behind the implementation of AI lies the crucial task of building a comprehensive full-stack AI cloud infrastructure.

"Full-stack" is a professional term originating from the computer field. In the AI era, the computing architecture has evolved from the traditional "chip-operating system-application" to a four-layer architecture of "chip-framework-model-application".

Taking Baidu as an example, it is one of the few domestic technology companies with full-chain self-research capabilities, covering everything from underlying chips, deep learning frameworks, to large language models, and further to industry and C-end applications. The biggest advantage of full-stack self-research is the ability to conduct end-to-end optimization, making AI systems more efficient, stable, and secure, while also lowering the threshold and cost for enterprises. Specifically, at the cloud computing level, AI cloud vendors need to build a complete AI cloud infrastructure ranging from computing platforms, data service platforms, model development platforms, to deployment platforms. A full-stack architecture can support enterprises' full-process needs from model training to business access. At the same time, possessing full-stack capabilities also means being able to serve diverse needs and adapt to complex scenarios. For example, in traditional manufacturing, finance, government affairs, and other industries with low data quality and limited budgets, full-stack AI cloud vendors can complete model fine-tuning and deployment at lower costs. In state-owned and central enterprises with extremely high requirements for security and autonomy, full-stack AI cloud vendors are the only option to complete the closed loop from hardware to delivery.

The key to AI implementation lies in full-stack infrastructure, and Alibaba Cloud, another heavyweight player in AI cloud, shares this view. The AI era places higher demands on the performance and efficiency of infrastructure, with the CPU-dominated computing system rapidly shifting to the GPU-dominated AI computing system. From building the "CIPU+Tianji" infrastructure in the CPU cloud era to focusing on AI in the GPU era, fully reconstructing underlying hardware, computing, storage, networking, databases, and data to accelerate model development and application, Alibaba Cloud is also actively building the strongest AI infrastructure for the AI era.

Final Thoughts

AI is reaching a pivotal point in its transition from "technological breakthrough" to "engineering delivery," and the competitive landscape within China's AI cloud sector is swiftly converging. The crux of this race in the future will revert to the fundamental question: who can effectively implement AI and deliver it to customers. In China, the leading players poised to ultimately dominate this race currently seem confined to Alibaba Cloud, Baidu Intelligent Cloud, and a select few others. This is a marathon, and the victor will be decided by who demonstrates the strongest strategic resolve.