The three major telecom operators have launched a battle for computing power positioning, aiming at monetizing AI applications

With China Mobile and China Unicom successively launching centralized procurement for computing power servers, the three major operators have started a positioning battle in the computing power field. Data shows that the existing intelligent computing scale of the three major operators has exceeded 100 EFLOPS (exaflops per second), covering multiple key node cities in the country.

Currently, the three major operators are accelerating the application of AI in fields such as mining, agriculture, inspection, healthcare, and urban governance, promoting business transformation and revenue growth through the deployment of large models and the introduction of AI intelligent entities.

In fact, in the first half of this year, both China Mobile and China Telecom achieved AI revenues in the tens of billions of yuan, making the monetization of AI applications a key factor in determining the success of their transformation. Industry insiders interviewed believe that in terms of AI infrastructure, it is not difficult for operators to achieve monetization; however, in areas such as large models, data, and even AI intelligent entities, attention should be paid to the integration and monetization of AI and network capabilities.

Positioning Quality Computing Power Resources

Recently, the bidding results for China Mobile's centralized procurement project for general-purpose artificial intelligence computing devices (inference type) for 2025 to 2026 have been announced. It is reported that the total procurement amount exceeds 5 billion yuan, divided into 6 bidding packages. Among them, bidding packages 1 to 4 are all CUDA ecosystem devices, with a total tax-inclusive price exceeding 1.7 billion yuan; bidding packages 5 and 6 are CANN ecosystem devices, with a total tax-inclusive price of about 3.4 billion yuan.

The bidding results show that the first four bidding packages were won by three traditional server manufacturers: ZTE, H3C, and Inspur; the last two packages were won by several Ascend computing power manufacturers, including Henan Kunlun, Huakun Zhenyu, and Yangtze Computing.

Rough estimates indicate that ZTE's winning bid amount in this procurement is about 885 million yuan, H3C's winning bid amount is about 548 million yuan, and Inspur's winning bid amount is about 297 million yuan; Henan Kunlun's winning bid amount is about 837 million yuan, Huakun Zhenyu's winning bid amount is about 668 million yuan, and Yangtze Computing's winning bid amount is about 597 million yuan.

Coincidentally, China Unicom also recently launched a centralized procurement project for general-purpose servers for 2025, with a total budget of up to 7.96 billion yuan (excluding tax), totaling 87,000 general-purpose servers, of which domestic computing power servers account for more than 90%.

Earlier this year, the three major operators announced their respective capital expenditure plans, among which China Unicom is expected to invest about 55 billion yuan in fixed assets in 2025, with computing power investment increasing by 28% year-on-year; China Mobile expects total capital expenditure of about 151.2 billion yuan for the year, with 37.3 billion yuan invested in computing power; China Telecom expects total capital expenditure to decline by 10.6% to 83.6 billion yuan, but computing power capital expenditure is expected to increase by 22% year-on-year. It is evident that all three major operators have prioritized computing power in their capital expenditures From the half-year reports disclosed by the three major telecom operators, China Mobile's self-built intelligent computing power has reached 33.3 EFLOPS. Meanwhile, China Mobile's two large-scale intelligent computing centers in Hohhot and Harbin are operating efficiently, and the globally leading "Computing Network Brain" has been implemented in multiple national nodes such as Wuhu, with a total intelligent computing scale reaching 61.3 EFLOPS.

China Telecom, based on the WanKa clusters in the Beijing-Tianjin-Hebei and Yangtze River Delta regions, is constructing super-node clusters in the Guangdong-Hong Kong-Macao Greater Bay Area hub, achieving a dual leap in both the quantity and quality of intelligent computing supply, with its own intelligent computing power reaching 43 EFLOPS.

China Unicom is building and operating WanKa intelligent computing centers in Shanghai Lingang, Hohhot, Ningxia Zhongwei, and Qinghai Sanjiangyuan, with a total intelligent computing scale reaching 30 EFLOPS.

It is worth mentioning that although half of the first half of the year has passed, the capital expenditure completion ratio of the three major operators is generally not high. For example, China Mobile completed capital expenditures of 58.4 billion yuan in the first half of the year, accounting for only 38.6% of the annual plan.

China Mobile's General Manager He Biao stated that the first half of the year was a phase for project research, verification, reserve, and promotion, resulting in a slower investment pace, but it will accelerate in the second half of the year. He further mentioned that the core of computing power is AI. Last year, China Mobile's AI investment exceeded 10 billion yuan, and this year, AI-related investments will still increase based on the original foundation, strengthening AI as the company's core strategic goal, forming new growth momentum, and creating a new growth curve.

High-Value AI Scenarios Emerge

Industry insiders introduced the main logic behind telecom operators' efforts in AI business to reporters: "First, layout from the underlying technology of AI, such as computing power base and large models; second, accelerate industrial implementation, such as AI applications in mining, farmland, inspection, drones, and robots."

The reporter noted that not long ago, the State-owned Assets Supervision and Administration Commission released the first batch of 40 high-value strategic AI scenarios for central enterprises, with all three major telecom operators included. Among them, China Unicom's selected scenario is "Intelligent Quality Control Auxiliary Diagnosis and Treatment for Medical Imaging and Testing Reports," China Telecom's selected scenario is "Full-level Intelligent Penetrating Supervision System," and China Mobile's selected scenario is "AI+ Processing of Hundreds of Millions of Cloud Threats in Public Cloud."

Regarding the aforementioned selected high-value scenarios, relevant personnel from China Mobile introduced to reporters that after AI empowerment, the cloud-based threat disposal capability has greatly improved, with the average processing time for security incident work orders reduced by 82.5%, the automation disposal rate of security alerts reaching 99%, and the false alarm rate reduced to 0.2%, saving over 10 million yuan in labor costs, helping Mobile Cloud achieve "zero" score loss in national offensive and defensive drills for five consecutive years.

"Medical AI is by no means a simple introduction of intelligent tools, but rather a revolutionary reconstruction of the traditional medical model," a relevant person from China Unicom told reporters, adding that China Unicom has launched a multi-dimensional layout in AI smart healthcare.

For example, China Unicom has assisted Beijing Friendship Hospital in deploying a full-version DeepSeek to meet the needs of different clinical and research usage scenarios, with the access volume of internal medicine knowledge Q&A accounting for over 50% Reporters also learned that in some cities, Chinese telecom operators are leveraging large-scale hydrological models to apply technologies such as AI and the Internet of Things in urban flood prevention. An operator representative told reporters that these AI applications not only visually display the operational status of the city's drainage system but also use advanced flood analysis models to predict which areas may experience overflow or flooding risks in the next two hours, achieving "flood prevention in advance."

From the information disclosed by the three major operators, in the first half of this year, China Mobile's AI + DICT signed projects reached 1,485, and it collaborated with central enterprises in energy, water conservancy, agriculture, and other industries to build large industry models; China Telecom has also developed over 160 AI applications around 21 key areas, effectively promoting cost reduction and quality improvement for enterprises; China Unicom has focused on key areas such as industrial manufacturing, economic operation, urban governance, and healthcare, with initial results in the large-scale commercialization of artificial intelligence applications.

Accelerating AI Integration and Monetization

"All three major operators are building their own intelligent computing centers and have their own large models," said Xiang Ligang, chairman of the Zhongguancun Information Consumption Alliance, in an interview with reporters. He mentioned that Chinese telecom operators have advantages in laying out AI businesses, as they can integrate their own storage and network capabilities and also rent out intelligent computing centers, with good growth in computing power leasing business.

China Mobile Chairman Yang Jie previously mentioned that in the first half of the year, China Mobile's direct and related income from AI reached tens of billions of yuan. He roughly categorized AI income into four major segments, among which AI computing power and intelligent computing services are important sources of revenue.

In Xiang Ligang's view, this year, due to the emergence of DeepSeek, the operators' computing power leasing business has achieved considerable growth. "In the future, the demand from enterprises for computing inference will be enormous, but not all will choose to build their own intelligent computing centers due to high construction costs; many small and medium-sized enterprises will consider renting computing resources from telecom operators."

He further mentioned that AI income based on computing power infrastructure can naturally integrate into the operators' regular business. For example, when providing intelligent computing and storage services to government and enterprise users, it must rely on its own network and bandwidth capabilities.

Reporters noted that relying on the "Jiutian" artificial intelligence large model matrix, China Mobile has achieved good returns. For instance, several central enterprises, including China National Petroleum Corporation, have built their large models based on the Jiutian foundation. Additionally, data processing has also contributed to revenue, with various AI applications targeting individuals, families, and government enterprises forming a scale.

Yang Jie predicts that as the industry's demand for high-quality data increases, the potential for AI income will continue to be released. In fact, the three major operators are continuously increasing their investment in data services. For example, China Unicom not only undertook the construction of a trusted data space in the marine field recommended by the Ministry of Natural Resources but also built high-quality datasets in the fields of information communication and healthcare.

However, it should also be noted that the proportion of AI income in the overall revenue of operators is still low. In the first half of this year, China Telecom achieved intelligent income of 6.3 billion yuan, a year-on-year increase of 89.4%, while China Unicom only stated that AI business has begun to scale monetization, becoming a new growth point "Some companies are willing to collaborate with telecom operators to develop large models, but the operators' business capabilities are essentially storage, intelligent computing, and networks." Xiang Ligang believes that from mobile communication to AI computing power, the core of operators remains information infrastructure. From cloud services to computing power services, the operators' advantage lies in the integration of cloud and network, as well as computing network.

Xiang Ligang believes that the new business of telecom operators must be integrated into traditional communication capabilities, which is also where operators have the strongest strength. Therefore, the purpose of operators focusing on AI business is to amplify the advantages of communication capabilities. Of course, the strength of AI business also requires support from communication capabilities.

Securities Times, original title: "The Three Major Operators Launch a Computing Power Positioning Battle Aiming at AI Application Monetization"

Risk Warning and Disclaimer

The market has risks, and investment should be cautious. This article does not constitute personal investment advice and does not take into account the specific investment goals, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Investment based on this is at one's own risk