Source: Financial Story Collection, Author: Jia Guipeng, Editor: Wan Tiannan

On the evening of August 10th, Alibaba released its financial report for the first quarter of the 2024 fiscal year. Among them, the revenue of Alibaba Cloud business increased by 4% year-on-year, and the adjusted EBITA profit surged by 106% year-on-year to 387 million yuan.

Although the growth rate is not impressive compared to other business sectors, Alibaba Cloud has managed to halt the previous downward trend in growth, which is not easy.

It is worth noting that compared to a few years ago when they were "advancing by leaps and bounds," internet cloud providers such as Alibaba Cloud, Tencent Cloud, and Baidu Cloud are all experiencing a "painful period" of slowing growth or even decline.

The "China Public Cloud Service Market (2022 H2) Tracking" report released by IDC shows that the overall market size of public cloud services in China (IaaS/PaaS/SaaS) reached 18.84 billion US dollars in the second half of 2022, with the IaaS market growing by 15.7% year-on-year and the PaaS market growing by 31.8% year-on-year.

In the IaaS+PaaS market, Alibaba Cloud's share decreased from 36.7% in the same period last year to 31.9%; Tencent Cloud's market share decreased from 11.1% to 9.9%. At the same time, state-owned cloud providers experienced fierce growth, with Tianyi Cloud, Mobile Cloud, and Unicom Cloud all more than doubling their year-on-year growth in 2022.

Amidst the overall slowdown, the cloud market is becoming more competitive, with the national advancement and private retreat of cloud computing, as well as the emergence of large AI models, bringing new changes to the cloud market.

Internet cloud providers are facing growth anxiety

Currently, cost reduction and efficiency improvement have become the main theme for many enterprises, and external expenditures, including those in the cloud service sector, are being reduced. This has also affected the revenue of internet cloud providers.

On the evening of August 10th, Alibaba Group released its financial report for the first quarter of the 2024 fiscal year. The report shows that the revenue of the Alibaba Cloud Intelligence Group increased by 4% year-on-year to 25.123 billion yuan.

Although returning to the growth track, looking at the longer term, Alibaba Cloud's growth rate has shown a general slowdown. From the 2019 fiscal year to the 2023 fiscal year, Alibaba Cloud's revenue growth rates were 84%, 62%, 50%, 23%, and 3.5%, respectively. Although Alibaba Cloud's revenue growth this fiscal quarter increased by 4%, higher than the 3.5% in the 2023 fiscal year, it still lags behind the previous years by a large margin.

As for Tencent Cloud, since 2014, it has become the second largest cloud service provider in China, second only to Alibaba Cloud. Its market share reached 10.3% in 2017, and even Huawei Cloud's market share did not exceed 1% that year.

Now, according to the data for the second half of 2022 released by IDC for the IaaS and IaaS+PaaS markets, Tencent Cloud has fallen to fourth place. It has been surpassed not only by Huawei but also by China Telecom Tianyi Cloud.

In Baidu's first quarter financial report, Baidu Intelligent Cloud achieved profitability (non-GAAP) for the first time in this quarter, with revenue increasing by 8% to 4.2 billion yuan. Although the growth rate has accelerated compared to the previous quarter's 4%, it is still much slower than the previous growth rate of over 20%.

Meanwhile, some competitors are "soaring," especially in the growth momentum of operator clouds.

In 2022, Tianyi Cloud's revenue was 57.9 billion yuan, a year-on-year increase of 107.5%; Mobile Cloud's revenue was 50.3 billion yuan, a year-on-year increase of 108.1%; Unicom Cloud's revenue was 36.1 billion yuan, a year-on-year increase of 121%. China Telecom Chairman Ke Ruiwen stated at the performance briefing that Tianyi Cloud's target for 2023 is to reach 100 billion yuan.

The even greater challenge is that although the Chinese cloud market will continue to grow for a considerable period of time, the competitive environment is becoming increasingly complex.

In particular, the construction of national computing power hubs and ten national data center clusters in the Beijing-Tianjin-Hebei region, the Yangtze River Delta, the Guangdong-Hong Kong-Macao Greater Bay Area, and six western provinces initiated by the National Development and Reform Commission and other departments last year will completely change China's current computing power structure. By then, cloud providers with state-owned backgrounds such as Inspur Cloud, Sugon Cloud, and China Electronics Cloud will compete for more market share.

The end of the internet dividend, customers turning into competitors

In fact, the rapid growth of internet cloud providers in the past was largely due to the massive migration of internet enterprises to the cloud. However, as the number of internet enterprises migrating to the cloud rapidly increases, the dividend in this area is approaching its peak, and non-internet customers have become the new growth drivers in the cloud market.

Alibaba Cloud, for example, saw a 3% year-on-year growth in the non-internet industry in Q4 2022, driving its revenue growth. According to Alibaba Cloud, the revenue from non-internet industry customers accounted for over 50% of Alibaba Cloud's total revenue at the beginning of 2022, and by the third quarter of last year, it had approached 60%.

Meanwhile, some of the internet giants that have contributed significantly to cloud computing revenue in the past are beginning to shift from "outsourcing" to "self-building" in the long term, which can save costs.

Kuaishou is one such example. Previously, Kuaishou was an important customer of Tencent Cloud.

Data shows that from December 31, 2020, to June 30, 2021, Kuaishou paid cloud service fees and technical fees to Tencent Cloud totaling 1.56 billion yuan and 1.11 billion yuan, respectively.

However, large platforms like Kuaishou, with a large user base, are offering increasingly diverse services and businesses, leading to a surge in demand for cloud services. This has resulted in higher cloud costs, and with the increasing amount of sensitive information such as transaction data and personal information from merchants and users, the security requirements are also increasing.

Therefore, self-built data centers have become a common option for many internet companies.

Against this backdrop, in 2021, Kuaishou invested 10 billion yuan to build a data center in Chahar High-Tech Development Zone in Ulanqab City. It was officially put into operation on July 28, 2023. After the completion of the Kuaishou data center, although it is not possible to completely cut off cooperation with Tencent Cloud, it does not prevent some of its business from being diverted to self-built data centers.

During the data center construction, Kuaishou also launched the StreamLake video cloud brand on August 10, 2022, in a productized form, providing a one-stop audio and video + AI solution. This brand is also Kuaishou's entry into the cloud service track and carries Kuaishou's ambition for "technology to business."

This strategy by Kuaishou not only gradually distances itself from Tencent Cloud but also puts it in competition with Tencent Cloud to seize market share in the cloud market.

In addition to Kuaishou, ByteDance, which also focuses on short video apps as its core business, announced its entry into the cloud track at an earlier time. In June 2021, ByteDance launched the cloud service brand Volcano Engine, providing digital services to enterprises. At the end of the same year, in the new round of organizational restructuring at ByteDance, Feishu and Volcano Engine became business lines at the same level as Douyin, Dali Education, games, and TikTok, indicating its strategic position has been further elevated.

Currently, Volcano Engine also serves many well-known external enterprises, including Geely Auto, Ideal Auto, China Merchants Bank, Yili, Pinduoduo, Taikang, and Unilever, among leading enterprises in various industries.

We can see that many internet enterprises that previously used Alibaba Cloud and Tencent Cloud are now building their own data centers, not only reducing their cooperation with leading internet cloud providers but also expanding their own businesses to seize market share in the cloud market, causing a diversion of market share.

Faced with the increasingly intense market competition, leading cloud computing providers such as Alibaba Cloud and Tencent Cloud have announced price reduction notices for their products. However, from the market's perspective, the "price war" is only a short-term tool for the manufacturers' game, and the broader government and enterprise market may be the key for internet cloud providers to increase their market share and regain growth.

"Loss" of government and enterprise business

Government and state-owned enterprise customers have always had a strong demand for cloud services, and they also have sufficient payment capabilities and willingness to make repeat purchases, with a relatively stable customer base.

The "In-Depth Research and Development Trend Report on China's Government Cloud Industry Market from 2023 to 2028" by the China Research Institute shows that out of the 334 prefecture-level administrative regions in China, 235 have already built or are bidding to build government clouds, accounting for possibly over 70% of the total. Looking ahead, the upgrade of local government clouds will bring new growth opportunities.

According to the "2022-2023 China Government Cloud Market Research Annual Report" released by CCID Consulting, the market size of China's government cloud reached 93.25 billion yuan in 2022. This figure already accounts for one-fifth of China's cloud computing scale of 455 billion yuan in 2022 (source: "Cloud Computing White Paper" released by the China Academy of Information and Communications Technology).

In the government and enterprise field, operators seem to be more "at home" than internet cloud providers.

According to the IDC report "China Government Cloud Market Share, 2022," in the public government cloud market, China Telecom ranks first, China Mobile ranks fourth, and China Unicom ranks sixth; in the government-exclusive cloud market, China Telecom, China Mobile, and China Unicom rank fourth, fifth, and sixth, respectively; in the government cloud operation and service market, China Mobile and China Telecom rank second and third.

Analysts revealed that the success of operators in the government and enterprise market is due to multiple factors. First, operators rely on their state-owned background, giving them a natural advantage in obtaining digital projects from government and enterprise clients. Second, their significant increase in market share is also closely related to their organizational restructuring and statistical methods. Third, it is related to the positioning of operator clouds. Finally, operators have unique advantages in reaching down to the grassroots level.

Currently, many smart city projects at the county level are won by local branches of operators.

As the national team of cloud services, the government and enterprise market is also a strong suit for Tianyi Cloud. Data shows that as of June 2023, Tianyi Cloud has undertaken over 20 provincial government cloud projects, over 300 municipal government cloud projects, and participated in over a thousand smart city project constructions, serving over 2 million customers; it has built 15 provincial and regional education cloud platforms, serving 350,000 schools; and it has built 12 provincial medical cloud platforms, with contracts for over 420 projects.

At the same time, internet cloud providers cannot undertake government and enterprise projects without cooperation with operators. This is because operators, as central enterprises, will bring cloud providers to take orders and then subcontract to cloud providers. They used to work on network construction with communication equipment companies, and now they are also working on government and enterprise digitalization with other cloud providers.

In this context, Tencent Cloud, Alibaba Cloud, and other internet cloud providers are emphasizing an integrated strategy. This also means that cooperation between operator clouds and internet giants will be greater than competition. Although operators are taking the lead in winning projects, in many winning projects of smart cities, they can also be seen cooperating with Tencent Cloud and Alibaba Cloud providers.

Winning the government and enterprise cloud market is also a consensus among internet cloud providers.

Since last year, Tencent Cloud has established a separate department specifically responsible for planning and expansion in the operator industry, aiming to strengthen cooperation with operators and achieve complementary advantages.

At the end of 2022, Ma Huateng stated at an internal staff meeting at Tencent that they should fully support the strategic shift from being an integrator to developing self-developed products.

On July 18, 2023, Tencent Cloud and China Unicom reached a deep cooperation in the Xi'an Digital Government project, jointly participating in the digital industrial construction in Xi'an.

Alibaba Cloud has also initiated multi-dimensional cooperation with operators, from core technology at the lowest level to business models and regional projects.

An insider at Alibaba Cloud revealed that they have dedicated teams for operators internally. "We ensure that our cooperation with operators is continuous and reliable."

At the Alibaba Cloud Summit in April, Alibaba Cloud announced its collaboration with China Mobile to build the Beijing Health Cloud, providing unified cloud service resources for medical and health institutions in Beijing to improve service efficiency and reduce medical costs for the public.

In government and enterprise construction projects, the actions of internet cloud providers go beyond cooperation with operators.

In recent years, the government and enterprise market has been repeatedly emphasized by Alibaba Cloud as a direction for heavy investment. At the Alibaba Cloud Summit, Xu Shijun, President of Alibaba Cloud's Intelligent Digital Government Industry, stated, "With the help of digital technology and means, we hope to quickly and systematically output Alibaba Cloud's previous digital capabilities and successful experiences to help the government efficiently fulfill its duties and serve the people."

Furthermore, in March 2022, Alibaba Cloud appointed Cai Yinghua, who was previously responsible for strategic planning and business promotion of Huawei's government and enterprise business in China, as Senior Vice President of Alibaba Group, hoping to leverage his experience and capabilities to help Alibaba Cloud expand into the government and enterprise market.

Under Cai Yinghua's leadership, Alibaba Cloud has reached cooperation with the Fengxian District People's Government in Shanghai, Shandong Port Group Co., Ltd., BBMG Group, Guotai Junan, and others.

Tencent also continues to focus on the government and enterprise field. At the 2022 Tencent Digital Ecology Conference, Li Qiang, Vice President of Tencent Group and President of Tencent's Government and Enterprise Business, mentioned that in the cloud field, the internet market is becoming saturated and is a stock market, so finding new growth points requires more attention to industries beyond the internet.

Tencent has made two key layouts in the government and enterprise business. One is for industries that are still in the early stages of digitalization but have strategic significance, such as the energy and industrial sectors corresponding to the carbon neutrality strategy. The other is for industries that can be connected to the C-end based on Tencent's existing advantages, such as cultural tourism, sports, agriculture, and other industries.

For example, in the energy industry, Tencent Cloud has launched the Ecological Aggregation Platform "Neng Carbon Workshop" and co-created scenario-based energy solutions with ecological partners. Data shows that Tencent Cloud currently serves over 20 central state-owned enterprises in the energy field.

Baidu is also catching up. Baidu's founder, chairman, and CEO Robin Li stated at the World Artificial Intelligence Conference in September 2022 that Baidu Intelligent Cloud continues to benefit from the growth opportunities of China's digital and intelligent transformation, supporting Baidu's customers in the manufacturing, transportation, energy, utilities, and public sectors with Baidu's AI capabilities.

Baidu Intelligent Cloud, which focuses on intelligent transportation, intelligent urban planning, and energy and power, currently provides smart scenario application solutions for over 70 cities nationwide, serving over 200 government customers.

However, despite the efforts of internet cloud providers to enter the government and enterprise market using various technologies and resources, the "China Government Cloud Market Share, 2022" report released by IDC shows that Alibaba Cloud ranks third in the public government cloud market, while Tencent Cloud ranks seventh. They did not make the list in the larger government-exclusive cloud and government cloud operation and service markets. This has inevitably left the long-standing leading internet cloud providers feeling "disheartened."

We believe that the initial cloud business model of internet cloud providers was to sell standardized services to the internet, e-commerce, developers, and startups. This business model, which can be quickly replicated and harvested, has led them to be more efficient with an internet mindset, while their service capabilities are weak and need to be strengthened.

Previously, "Financial Story Collection" interviewed a startup team using products from internet cloud providers. The team had some technical issues they did not fully understand and wanted to get support from the provider, but after contacting customer service, they were given the contact information for sales, leaving the startup team feeling "helpless and amused."

Unlike internet companies, government and enterprise users require solutions tailored to the business scenarios, processes, information planning, pain points, and needs of the cloud providers. This places demands on the organizational structure and integrated service capabilities of cloud providers, such as Huawei, which has established a precise coupling organizational structure with government and enterprise major customers. It seems that internet cloud providers have not been able to make this transition.

At a deeper level, the government and enterprise market requires players to have greater differentiation. For government and enterprise users, service is the key element for cloud providers to remove the label of homogeneity and is the pivot for various cloud providers to create distance in the future. Currently, the competition among cloud providers regarding services has become apparent.

Therefore, in the short term, it is difficult to quickly develop the government and enterprise market, and finding a completely new blue ocean market has become another way for internet cloud providers to resist the decline in cloud market share.

Ambitions to Rise with AI

Although internet cloud providers seem to have lost some of the limelight to operators in the government and enterprise market, they have more chances of winning in the upcoming wave of intelligent clouds.

Many industry analysts believe that generative AI and the AI ecosystem behind it will completely change the industry rules of cloud computing, giving rise to new market opportunities.

According to the McKinsey report "The Economic Potential of Generative AI," generative AI can improve the production efficiency of 2,100 specific jobs in 850 occupations, with a potential cost reduction of up to $6.1 to $7.9 trillion. For enterprises, it can be applied to solve specific business challenges in 63 scenarios across 16 business departments, bringing a value of $2.6 to $4.4 trillion to enterprises.

The Goldman Sachs report also shows that generative AI can drive global GDP growth by 7%, nearly $7 trillion.

From an application perspective, the currently popular chatbots are just a part of generative AI. AI assistants, intelligent customer service, and personalized recommendations are already being implemented through artificial intelligence. Knowledge base search, meeting minutes, text summaries, content or code creation based on generative AI will elevate production efficiency to a new level, making it a reality to create an app or write a script in minutes.

As the engine and technical infrastructure for generative AI, AI large models are becoming the target of enterprise pursuit. Currently, Baidu Wenxin Yiyuan, Huawei Pangu, 360 Zhineng Nao, Sensetime Ririxin, Alibaba Tongyi Qianwen, JD Lingxi, Kunlun Wanwei Tiangong, iFlytek Xinghuo, and Tencent Hunyuan, among others, have successively made their debut, creating a "thousand-model war" as hyped by the public.

Faced with revolutionary production scenarios and updates to production modes, more enterprises will turn their attention to the AI market. However, deploying and using AI services in actual business scenarios presents challenges such as limited computing resources, poor data quality, high investment costs, and lack of professional experience. This has led more enterprises to adopt Model as a Service (MaaS) from cloud providers as a solution for deploying large model services.

Currently, as large models mature, the cloud computing service landscape is gradually transitioning from a platform based on computing power (IaaS) to a platform based on model capabilities (MaaS). MaaS learns from massive and diverse scenario data, summarizes the common features and rules of different scenario businesses, and becomes a model base with generalization capabilities. The data set warehouse, model warehouse, and computing platform provide zero-threshold model experiences, quick model usage, complete custom model chains, and cloud-based model deployment to each user.

In form, MaaS is a typical combination of cloud computing and artificial intelligence, integrating cloud services and artificial intelligence innovation to standardize products such as computing power, frameworks, models, and scenario applications, thereby lowering the threshold for enterprise users to obtain and use artificial intelligence and empowering innovation and cost reduction in various aspects.

In May of this year, Baidu CTO Wang Haifeng stated that MaaS will become the mainstream business model of cloud computing in the future, and various applications will be developed based on large models. Each industry needs to create its own large model.

After the release of "Wenxin Yiyuan," Baidu quietly announced the progress of large model landing for enterprise clients. According to reports, Baidu has 11 large models covering key areas such as transportation, energy, and has the largest industrial application scale in China.

In an interview in May, Baidu's Vice President Hou Zhenyu stated that after the release of "Wenxin Yiyuan," Baidu Cloud has attracted more new customers. On the one hand, customers who were not previously on Baidu Cloud have started to explore business cooperation with Baidu Cloud, and on the other hand, customers are prioritizing Baidu Cloud when allocating budgets.

Additionally, Baidu continues to attract users through various means. On August 2, Baidu's AI large model platform, Wenxin Qianfan, announced the integration of the entire series of open-source models LLaMA2, and the platform's available large models have increased to 33.

A model developer told "Financial Story Collection" that the main reason for open-sourcing large models at this stage is to gain market share for free. For many basic users, no amount of marketing can compare to the affordability of open-source models.

Similarly, in May, Alibaba Cloud CTO Zhou Jingren stated at a meeting in Zhongguancun that the concept of MaaS is being widely accepted, and models will become an important production element for business and development systems.

In April of this year, Alibaba Cloud released the AI large model "Tongyi Qianwen" and announced that it will be integrated into all Alibaba products for comprehensive upgrades. Currently, more than 12 product lines and over 40 scenarios, including DingTalk's knowledge base, forms, mind maps, whiteboards, Teambition, and email, have been integrated with large models.

To accelerate the landing of large models in various industries, Alibaba Cloud has also launched the "Tongyi Qianwen Partner Program," which will provide technical, service, and product support in the large model field to partners, promoting the application of large models in different industries. The first batch of partners covers the oil and gas, power, transportation, finance, tourism, enterprise services, and communication industries.

On August 10, Alibaba Group Chairman and CEO, Alibaba Cloud Intelligence Group Chairman and CEO Zhang Yong stated at the financial report meeting, "The industry has a strong demand for AI cloud services, but due to short-term supply chain challenges, this demand has only been partially met. The incremental opportunities brought by AI-related services are just beginning to be unleashed. This means that Alibaba Cloud will benefit from this in the long term."

Furthermore, Zhang Yong also mentioned that Alibaba Cloud will continue to advance the development of self-developed large models and then open-source the large models, allowing more people to use this basic model on Alibaba Cloud.

On August 3rd, Alibaba Cloud announced the open-source release of its self-developed large model "Tongyi Qianwen." Specifically, the general model Qwen-7B with 7 billion parameters and the dialogue model Qwen-7B-Chat are both available on the AI model community ModelScope, open-source, free, and available for commercial use.

After the open-source release of the Tongyi Qianwen large model, it can help users simplify the process of model training and deployment. Users do not need to train the model from scratch; they only need to download the pre-trained model and fine-tune it to quickly build high-quality models. This will increase the stickiness of users adopting the pre-trained models of Tongyi Qianwen for their products.

As of the end of July, Alibaba Cloud's ModelScope community has gathered over 2 million AI developers and more than 1,000 high-quality AI models, with a total download volume of over 45 million. The contributors of the models cover core institutions in the domestic large model race.

In the large model field, Tencent has also been active. Its self-developed "Tencent Hunyuan Large Model" has entered the internal testing phase.

According to insiders at Tencent, the Hunyuan large model supports text input, voice input, and can generate content such as product recommendations for Xiaohongshu, short video scripts, and WeChat moments posts. It also supports AIGC painting generation in styles such as cyberpunk, pixel art, and mosaic.

At the same time, Tencent officially announced the industry large model research and development progress and the panoramic view of MaaS capabilities for the first time in June.

Unlike other major companies, Tencent Cloud's MaaS is a industry-specific model store built on an IT platform, covering 10 major industries including finance, cultural tourism, government affairs, media, and education. Based on basic models, customers only need to add their own scenario data to quickly generate exclusive models tailored to their business needs.

Tencent Cloud's Vice President and Head of Tencent Cloud Intelligence, Wu Yunsheng, mentioned at the 2023 World Artificial Intelligence Conference that compared to general large models, industry-specific large models are more concrete. When combined with Tencent's years of experience and data in the industrial internet, they can solve many industry problems in a different way.

Overall, using MaaS as a lever to change the way cloud providers offer services will greatly reduce the usage threshold for enterprise users. Cloud computing is transitioning from selling resources to selling capabilities and then to selling services, continuously breaking down the barriers to technology implementation. Whether for large enterprises or small and medium-sized businesses, MaaS is a better solution.

Large models will become an important variable in the next phase of competition among cloud providers.

Conclusion

In the domestic cloud computing market, as the internet dividend gradually reaches its peak, internet cloud providers are facing growth bottlenecks. Moreover, facing the rapid progress of operator clouds in the government and enterprise market, internet cloud providers are feeling anxious. This has prompted them to accelerate the expansion of the government and enterprise market while seizing the large model dividend to consolidate their position in the cloud market.

The battle of cloud computing has only reached halftime. In the future, all players will face more intense competition, more complex battles, and a more diverse test of capabilities.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。