On November 1, news of DeepSeek's latest funding round broke, revealing that two major telecom operators, China Mobile and China Telecom, appeared simultaneously on the investor list, planning to invest a combined total exceeding CNY 10 billion. Specifically, China Mobile plans to invest CNY 8 billion, setting a new record for the scale of investment by a telecom operator in a domestic large model company.
How should we interpret this move? In my view, the significance of this transaction extends far beyond a routine industrial investment. It is essentially a strategic positioning by state-owned capital for DeepSeek, an AI core asset of national importance. It also marks a new stage in the development of domestic large models, characterized by deep integration with industrial capital.
- 01 - The Strategic Positioning Behind the CNY 10 Billion Investment
The first thing many people focus on is the weight of the CNY 8 billion figure. From the perspective of the operators' own financials, this investment approaches China Mobile's full-year capital expenditure for intelligent networks in 2026. Even calculating China Telecom at its lower bound of CNY 2 billion, it exceeds the total size of its single AI industry fund. In terms of investment intensity alone, this is an order of magnitude higher than previous lead investments in the hundreds of millions.
However, viewed from another angle, the actual financial pressure of this investment is not significant. China Mobile's CNY 8 billion represents only 5.8% of its net profit for 2025, roughly equivalent to twenty-some days of profit. Even if China Telecom calculates based on the upper limit of CNY 8 billion, it remains less than one-third of last year's net profit. This indicates that both companies completed a strategic-level asset allocation within a fully controllable range.
More critical is the value of the target itself. DeepSeek is not an ordinary AI startup; it is one of the few domestic players capable of catching up with the global first tier in general-purpose large models and currently boasts one of the most active open-source large model ecosystems in China. Large models are the foundational infrastructure for the next generation of digital economy, akin to the power system in the industrial era and the operating system in the information age, serving as core tools supporting intelligent upgrades across all industries. Whether one can master autonomous and controllable general-purpose large models directly relates to the discourse power in future industrial intelligence, as well as data security and supply chain security in key sectors. From this dimension, leading domestic general-purpose large models have already become core digital assets with national-level value.
The operators' shift from wait-and-see trial waters to heavy betting essentially reflects their recognition of the strategic attributes of large models. They are not seeking short-term financial returns but rather aiming to secure a position at the core nodes of the AI industry, keeping the initiative over core digital assets in their own hands.
- 02 - Bidirectional Deep Integration: Aligning National Strategy with Industrial Needs
DeepSeek's current funding round is not lacking in capital. With nearly 30 investors and a financing scale reaching hundreds of billions, market recognition is evident. The fact that DeepSeek explicitly required business synergy and resource support from investors underscores that the core purpose of this round was to find industrial partners, not just financial investors.
For large model companies at their current stage of development, the biggest bottleneck is no longer just algorithm iteration. Stable and low-cost large-scale computing power supply, nationwide IDC coverage with accompanying power facilities, and implementation channels penetrating various industries are shortcomings difficult to bridge quickly through burning cash alone. These happen to be the core advantages of telecom operators.
Operators possess nationwide data center resources, natural advantages in backbone network bandwidth, and mature power and liquid cooling support systems, enabling them to rapidly accommodate the deployment needs of ultra-large-scale intelligent computing clusters. Simultaneously, operators have deeply cultivated the government and enterprise markets for years, possessing complete service systems and reach channels ranging from central enterprises to local governments, and from key industries to SMEs. After DeepSeek integrates deeply with operators, whether in computing power assurance for model training or advancing private deployment in specific industries, it will gain immense support.
A deeper point of alignment lies in the strategic direction of autonomy and controllability. DeepSeek has previously open-sourced Ascend adaptation components, actively promoting the implementation of domestic computing power routes, which aligns perfectly with the industrial direction of "training domestic models on domestic chips." As the main force of computing power within the state-owned capital system, investing in DeepSeek allows operators to essentially bind national-level computing infrastructure with core domestic large model technology, achieving end-to-end autonomy and controllability from chips to computing power to models.
This is the most core strategic significance of China Mobile's massive investment. It is not merely a corporate business layout but, acting as industrial capital, supports the autonomous development of domestic AI technology and strengthens the underlying foundation of the entire national AI industry. For key sectors such as government affairs, finance, and energy, intelligent upgrades must rely on secure and controllable domestic models. The combination of operators and domestic large models precisely meets compliance requirements and security needs in critical fields, supporting the smooth advancement of intelligent transformation in core industries. This is where the real-world value of national-level assets lies.
- 03 - Dual Track of R&D and Investment: Building an Autonomous and Controllable AI Ecosystem
In reality, this is not the first time operators have invested in large model startups. China Mobile has invested in Kimi, and China Telecom led the investment in Modelbest. Adding DeepSeek, both have established multi-target investment matrices. Meanwhile, China Mobile continues its self-developed Jiutian large model, and China Telecom maintains its Xingchen (Star) large model, never stopping their R&D efforts.
Many wonder why they continue to invest in external companies when they have their own models. The answer is clear: self-development secures the base, while investment expands the ecosystem; both tracks are indispensable.
The core role of self-developed large models is to guard the fundamental business base and ensure the autonomy and controllability of core services. Internally, they support cost reduction and efficiency gains in scenarios like network O&M, customer service, and internal management. Externally, they meet basic intelligent service needs. However, large model technology iterates extremely fast, and demands in niche scenarios vary widely. A single self-developed team cannot cover all technical directions and application areas.
By investing in multiple leading large model companies with distinct characteristics, operators can quickly complete their capability matrix, offering customers richer choices. More importantly, this layout drives the growth of the entire domestic AI ecosystem. By putting forward CNY 8 billion in real money, China Mobile did not just acquire equity in a company; it sent a clear signal to the entire domestic AI track: state-owned capital will continuously tilt towards top-tier enterprises with core technologies committed to autonomous paths. This guiding effect will attract more industrial resources and talent to the domestic autonomous route, pushing the entire ecosystem toward maturity.
- 04 - Accelerating Entry of State-Owned Capital: Domestic AI Enters a New Stage of Industrial Implementation
In this funding round, state-owned capital contributed approximately 27% of the total share, with the two operators being the most core contributors within the state-owned camp. This signal is very clear: state-owned capital is shifting from passive observation to active layout in the large model track, with resources continuing to concentrate on top-tier autonomous and controllable targets.
Prior to this, the main players in the large model track were internet companies and startups, with competition focused on parameter scale and algorithm iteration. As industrial capital represented by operators and state-owned capital enter, the underlying logic of competition is changing. Future battles will increasingly emphasize comprehensive capabilities across the entire industrial chain. Only those who can secure stable and reliable computing power supply, connect more implementation scenarios, and build a secure and autonomous industrial ecosystem will stand firm in long-term competition.
For the entire domestic AI industry, this is an important turning point. National-level core models, paired with national-level computing infrastructure and industrial channels, will significantly accelerate the transition of large models from technological breakthroughs to industrial scaling. When AI capabilities rapidly penetrate thousands of industries, supporting the nation's industrial digitalization and intelligent transformation, the value generated will far exceed the commercial returns of a single company.
China Mobile and China Telecom's investment of over CNY 10 billion in DeepSeek is an investment with distinct strategic attributes. It is a key step in the operators' own AI layout and a crucial measure for state-owned capital to strengthen the autonomous and controllable foundation of domestic AI. As a benchmark for domestic large models with national-level value, DeepSeek will further accelerate technological iteration and industrial implementation with the resource support from operators. The entire domestic AI industry will also walk a more solid path of autonomous development through this deep integration of industrial capital and core technology.