虎嗅

From traffic to computing power, China Mobile aims to turn AI into another ‘network’

原文:从流量到算力,中国移动想把AI再做成一张“网络”

Summary of Key Points

China Mobile experienced a decline in both revenue and net profit in the first half of 2026 (revenue decreased by 1.1%, net profit decreased by 6.3%). The growth of its traditional telecommunications business has hit a bottleneck: although users are consuming more data, the company is earning less from it. The company is shifting funds previously invested in 5G infrastructure to new areas such as computing power and AI data centers (AIDCs), attempting to replicate the successful model of "building a nationwide network and then selling standardized services." It aims to utilize its existing base stations, fiber optic networks, and government and enterprise customer relationships to establish a national computing power network. However, although the revenue from these new businesses is growing rapidly, whether their profitability and cash flow can match those of traditional telecommunications services remains to be seen.

I. Traditional Business: More Data Consumption, Less Profit

In the past, China Mobile made money by upgrading its network, attracting more users, and increasing data prices. This strategy no longer works:

  • Data Growth ≠ Revenue Growth: Data usage increased by 16% in the first half of 2026, but data revenue has been declining for three consecutive years (2025 saw a 6.8% decrease compared to the peak in 2021).
  • Declining User Payment Willingness: The average revenue per user (ARPU) dropped from 52.4 yuan in 2023 to 45.1 yuan in 2026, with ARPU for household users even surpassing that of individual users for the first time, indicating greater pressure on individual business segments.
  • Costs Stagnate Despite Increased Data Usage: The fixed costs of communication networks are high, but additional data usage hardly increases these costs (existing base stations can handle the demand), leaving little room for price increases and reducing revenue flexibility.

In short, as users consume more data, China Mobile's profits are declining, and it needs to find new ways to generate income.

II. Where to Invest the Money: Shifting from 5G to Computing Power

The peak of 5G construction has passed, and China Mobile is reallocationing its funds:

  • Change in Capital Expenditure: Investment in communication networks decreased by 20.3% in 2026, while investment in computing power networks increased by 62.4% (shift from building base stations to constructing intelligent computing centers and AIDCs).
  • Improved Cash Flow: Free cash flow increased by 111.6% in the first half of 2026, but this money needs to be invested in GPU and computing power management systems.
  • Challenges with New Assets: Whether these new assets will be profitable depends on factors such as utilization rates (a high utilization rate of over 90% is positive), depreciation costs (GPUs are updated frequently, leading to higher depreciation expenses), and payment collection speeds (new businesses often involve government and enterprise projects with slower payments).

In other words, funds are moving from old networks to new computing power infrastructure, but whether the investment will be profitable is still uncertain.

III. Replicating Success: Turning Computing Power into a "National Network"

China Mobile excels at building networks and selling standardized services. It is now applying this approach to AI:

  • Past vs. Present: Previously, it connected base stations and fiber optic networks to provide data services; now, it connects intelligent computing centers and AIDCs to offer computing power services (such as GPU capacity and data center resources).
  • Core Competence Remains the Same: Users do not need to worry about where the computing power is located (Harbin or Hohhot); they simply submit their requirements (e.g., amount of computing power needed, latency requirements), and China Mobile's "computing network brain" will handle the scheduling automatically.
  • Utilization of Existing Assets: The company's nationwide network of 6.86 million base stations and 35.86 million kilometers of fiber optic cables can be used as a high-speed infrastructure for data transmission, an advantage that internet companies do not have.

Essentially, China Mobile is acting as an "infrastructure operator" for AI, providing the necessary infrastructure without developing AI models or applications itself.

IV. Comparing with Alibaba and Tencent

All three companies are investing in AI, but their approaches differ significantly due to their core competencies:

  • Alibaba: Covers the entire AI value chain from chips to models to applications (e.g., Qianwen model, e-commerce AI) due to its strengths in cloud computing and e-commerce.
  • Tencent: Uses AI to enhance existing services (e.g., WeChat recommendations, game AI) leveraging its large user base.
  • China Mobile: Focusing on building computing power infrastructure and networks, utilizing its nationwide hardware network and government and enterprise customer base (over 36 million customers) to offer computing power as a service to governments and enterprises.

There is no clear winner; each company is focusing on areas where it has a competitive advantage.

V. Rapid Growth of New Businesses, but Profitability Uncertain

Revenues related to computing power have increased significantly (AIDC grew by 486%, intelligent computing services by 130%), but there are two concerns:

  • Profit Margins: AIDC generates stable revenue from renting data centers, while intelligent computing involves investing in GPUs and dealing with depreciation costs. China Mobile has not disclosed the profit margins for these areas, so high growth does not necessarily equate to high profitability.
  • Cash Flow Pressure: Accounts receivable increased from 75.7 billion yuan in 2024 to 122.2 billion yuan in the first half of 2026, mainly due to the slow payment collection associated with government and enterprise projects.

In conclusion, China Mobile has identified a viable entry point for AI, but whether it can achieve stable profitability like its telecommunications business depends on factors such as asset utilization, payment collection, and depreciation management.

Final Remark

China Mobile is transitioning from a company that sells data to one that sells computing power, leveraging its existing network infrastructure to pursue the future of AI. The success of this strategy will depend on whether new businesses can be as profitable and reliable as its traditional data services.