This marks the 1311th exclusive piece from 'Frontiers of New Energy.' For more insightful content, click 'Frontiers of New Energy' above to follow and 'star' our account. Please note, the views expressed in this article are solely those of 'Frontiers of New Energy' and should not be construed as investment advice. The author does not lead investment groups, offer paid stock tips, or handle client finances.

--------

Cloud computing undoubtedly stands as a groundbreaking commercial innovation. Business fundamentals suggest its continued growth alongside the internet industry, a trend mirrored in the real-world business landscape. However, cloud computing enterprises in China and the United States exhibit notable disparities, shaped by their respective national contexts.

 01 Divergent Cloud Computing Landscapes in China and the U.S.

In the United States, cloud computing is undeniably lucrative, as reflected in the gross margins of leading American cloud giants. Azure (Microsoft Cloud) boasts a gross margin as high as 40%, AWS (Amazon Cloud) stands at 35%, and even Google Cloud, with the lowest margin, reaches 30%. Google Cloud's lower margin is attributed to its ongoing production capacity expansion, with scale advantages yet to be fully leveraged. Ultimately, its gross margin is expected to align with AWS without much difficulty.

In contrast, domestic cloud computing companies face a different reality. Alibaba Cloud's gross margin is less than 10%, estimated to be around 8% to 9%. Tencent Cloud has just turned a profit, despite its revenue nearing 100 billion (based on the 2024 annual report, with cloud services and fintech businesses achieving profitable scale in 2025). Huawei Cloud is still incurring losses.

Why does the same business yield vastly different results?

This divergence stems from the varying development stages and scales of cloud service enterprises in China and the United States. However, the core reason is aptly summarized by a quote from 'The Annals of Yan Zi': 'An orange tree grown south of the Huai River remains an orange tree; grown north, it becomes a trifoliate orange. The leaves may appear similar, but the flavors differ.' Why? Because the soil and water conditions vary.

It all boils down to the environment!

The key distinction lies in the differing development stages, labor costs, and ultimately, the varying payment capacities or willingness of corporate clients to invest in cloud services between China and the United States.

Before delving deeper, let's briefly introduce some key cloud computing terms: IaaS, PaaS, and SaaS.

IaaS, or Infrastructure as a Service, involves cloud computing companies renting out hardware such as servers, data centers, bandwidth, and storage. Clients are responsible for tasks like system installation, software development, coding, and application building. Essentially, clients are just renting the hardware.

PaaS, or Platform as a Service, takes it a step further. Cloud computing companies provide development platforms, allowing clients to focus on application writing.

SaaS, or Software as a Service, offers ready-to-use software, eliminating the need for clients to worry about the underlying infrastructure. Common examples include CRM software like DingTalk and Feishu.

These services represent a progressive relationship. For cloud service providers, the ideal scenario is for clients to opt for SaaS services due to their high differentiation and greater pricing power. Services like IaaS, which involve hardware stacking, offer little differentiation, leading to homogeneous competition. As long as stability is ensured, the platforms developed by clients on your hardware are not significantly different. Therefore, cloud computing companies must invest heavily in self-developed critical hardware, such as GPUs and CPUs, as well as in building data centers, essentially striving to extract profits from the upstream.

Now, the crucial difference between Chinese and American cloud service companies becomes apparent. In the United States, the product sales structure of cloud service companies primarily consists of PaaS and SaaS, accounting for over 70%, while pure 'hardware leasing' (IaaS) accounts for less than 30% and serves as a loss leader. In contrast, the data is reversed in China, with IaaS being the dominant force and PaaS and SaaS having relatively low shares.

So, why is there such a significant difference in the product sales structure between Chinese and American cloud giants?

 02 Unveiling the Disparities in Cloud Computing Between China and the U.S.

As mentioned earlier, the differing development stages lead to varying labor costs and, ultimately, different payment capacities. In the United States, high labor costs make opting for IaaS to rent hardware and then hiring people to develop platforms and applications not significantly cheaper than directly renting PaaS and SaaS. Moreover, it introduces management complexities, making it simpler and more cost-effective to choose SaaS directly.

Conversely, in China, low labor costs prevail. With a vast workforce, there is no shortage of programmers. The primary clients of cloud computing are internet companies, which are well-stocked with programming talent. Instead of directly renting PaaS and SaaS, they can opt to rent hardware (IaaS) and let their programmers handle platform and application development, resulting in lower costs and better alignment with the company's business characteristics. Why wouldn't they choose this option?

China's rapid development over the past few decades has been fueled by its vast demographic dividend, whether in low- to mid-end manufacturing or the current engineer dividend. Essentially, it boils down to low labor costs.

Capital seeks returns. As long as labor costs are lower than using machines, there is little incentive to use machines. Similarly, as long as labor costs remain low, there is no motivation to use more expensive SaaS services. Instead, companies opt for IaaS and develop their own applications.

This challenging payment environment has forced domestic cloud service providers into fierce competition, resulting in many CRM vendors offering basic functions for free and charging low prices for some features. Since it's difficult to earn excess profits from clients, they can only compete internally, typically by focusing on scale and seeking scale advantages. The emphasis is on volume, trading price for quantity, effectively turning the high-tech cloud computing industry into a commodity-like business.

As the saying goes, 'An orange tree grown south of the Huai River remains an orange tree; grown north, it becomes a trifoliate orange.' The environment dictates this outcome, and domestic cloud service providers have no choice but to adapt. The development stage and national conditions are what they are.

 03 Alibaba Cloud's Business on the Brink of Transformation

If things were to continue as they are, Alibaba's cloud services would hold little promise. Even without considering competition, the profit ceiling would be locked in. Fortunately, the advent of the AI era has opened up new possibilities for cloud services, propelling cloud computing beyond IaaS, PaaS, and SaaS into MaaS, or Model as a Service. Cloud service companies now directly package trained large models into APIs for use. As these large models become increasingly powerful and versatile, they are poised to become genuine productivity tools, potentially marking a singular moment in the cloud service industry.

Compared to PaaS and SaaS, the cost of building a high-performance large model is prohibitive for many downstream vendors, let alone small and medium-sized enterprises (SMEs). Even giants like Meta and Tencent struggle to create a user-friendly large model.

For SMEs, as long as the capabilities of a large model meet their needs, there is no need to invest in building their own. Renting the model's capabilities suffices, especially since the model's capabilities can significantly reduce costs for the enterprise, far exceeding the rental costs. As long as the math adds up, enterprises will not hesitate to pay.

In many areas, this calculation already balances out, as evidenced by many internet companies laying off programmers. As the capabilities of large models continue to iterate, this process will persist and intensify, eliminating concerns about the math not adding up.

Traditional cloud service providers can then shift from primarily selling IaaS resources to offering high-value-added services such as MaaS, AI PaaS, intelligent computing clusters, and industry-specific large models. Cloud service providers will no longer be just 'server sellers' but 'AI capability sellers.' Their business models will transition from one-time charges to continuous payment models based on tokens, usage volume, or subscriptions, resulting in a more sustainable, stable, and healthy revenue structure.

Crucially, traditional cloud services require extensive customization development due to varying industry needs, even when selling PaaS and SaaS. This process is costly and difficult to price competitively. However, as the capabilities of large models continue to improve, they can eventually meet the needs of most industries and enterprises without extensive customization. MaaS has the potential to penetrate beyond the internet sector into the entire real economy, including finance, education, manufacturing, healthcare, and more. As demand grows, costs can increase, representing a significant leap in value.

This is likely why Alibaba is so optimistic about its cloud computing business in the coming years.

If the capabilities of large models can indeed continue to iterate and evolve to the point where they can meet the needs of most industries and enterprises with little to no customization, Alibaba's internal expectations for cloud computing could be realized, potentially even exceeded.

Of course, whether this will ultimately happen remains to be seen. However, given my optimistic view of AI technology iteration, I am full of anticipation. Based on the current synergy of cloud services and large model capabilities, Alibaba is indeed the company in China most likely to be the first to reap the benefits of this era, unparalleled in its potential.