Tencent has added fuel to the fire for domestic computing power.
```
Author | Huang Yu
Editor | Wang Xiaojuan
Over the past year, AI large model applications have been accelerating their implementation. The roaring wave of AI is generating global demand for computing power at an exponential rate. In this worldwide AI arms race, whoever can provide effective computing power can become one of the winners at the table today.
At Tencent's most important annual To B industry event—the Tencent Global Digital Ecosystem Summit—Tencent's deployment in computing power has also attracted wide attention.
Tencent Cloud announced that Tencent has now fully adapted to mainstream domestic chips. At the same time, holistic hardware-software co-optimization is a long-term strategic investment for Tencent Cloud, which provides cost-effective AI computing power by integrating different types of chips via a heterogeneous computing platform.
In other words, even with a focus on software, Tencent is also dedicated to building solid AI infrastructure. Meanwhile, against the background of an ongoing strengthening of domestic chip replacement, Tencent's comprehensive adaptation move undoubtedly further fuels the domestic computing power sector.
Of course, the AI large model competition is a marathon. Tencent's core advantage in AI lies in its wide range of applications. Thus, for a long period after the AI large model wave began, Tencent did not rush to seize the AI-related market. Instead, it prioritized empowering its self-developed large models and its massive own business.
After more than two years of effort, AI technology has become the core growth engine for Tencent and Tencent Cloud, with a positive cycle of AI investment and high-quality growth being continually reinforced. Against this backdrop, since the beginning of this year, Tencent has switched from a defensive to an offensive strategy in AI, with a more proactive and open Tencent emerging.
AI is also bringing a value reassessment to Tencent. Recently, Tencent's stock price hit a nearly 4-year high, and returning to a market capitalization peak of 7 trillion seems not far off. However, this is not the end. The market expects that AI can recreate Tencent and bring even greater possibilities.
Computing Power Revenue Growth
Three years ago, Tencent Cloud put forward "Better to have 150 pounds of muscle than 200 pounds of flab." After three years of transformation, focusing on self-development and driving integration, Tencent Cloud now stands tall within the entire Tencent Group.
Tang Daosheng, Senior Executive Vice President of Tencent Group and CEO of Cloud & Smart Industries Group, revealed that in Q2 2025, Tencent's To B revenue achieved double-digit growth reaching 55.5 billion yuan; Tencent Cloud's overseas customer scale doubled year-over-year in the past year.
"I can proudly say that after years of continuous focus on core products and technology, slimming down and building muscle, Tencent Cloud has become a highly competitive 'fit competitor' in the cloud market."
Undoubtedly, AI has become a strong driving force for Tencent Cloud's growth. In the second quarter of this year, Tencent Cloud's revenue growth accelerated compared to recent quarters. This was mainly due to increased enterprise demand for AI-related services, including GPU rental and API token usage, as well as merchants' technology service fees growth.
Computing power is indeed good business. Tang Daosheng told Wallstreetcn that the AI wave's boost for Tencent Cloud AI Infra infrastructure is undeniable, and in the future income generated by AI-related business, growth in AI infrastructure-related areas will become even more apparent.
Tencent's investment in computing power is also significant. According to Tencent's financial report, since the strategic acceleration in Q4 last year, Tencent's cumulative capital expenditure reached 83.16 billion yuan as of Q2 this year, mainly to support AI-related business development.
At the August results meeting, Tencent's management also said Tencent has enough chips to train and continuously upgrade existing models. In addition, Tencent is improving inference efficiency through a large amount of software-level optimization, so as to handle more workloads with the same number of chips.
Li Qiang, Vice President of Tencent Group and President of Government & Enterprise Business, pointed out that as a cloud computing vendor, under the "AI+" backdrop, the first step should be to build a larger and more abundant computing power foundation—infrastructure that will benefit thousands of industries. In the second half of this year, Tencent will continue to invest, building a large-scale AI Infra worldwide.
Fully adapting to domestic chips is also undoubtedly an important step for Tencent Cloud to enhance its AI Infra.
Tang Daosheng explained that Tencent is cooperating with many chip manufacturers for adaptation. Actually, there are many types of models, large and small. In different scenarios, chip configurations needed are different, so Tencent will keep cooperating with multiple vendors.
This means that while some tech giants are making their own chips, Tencent chooses to focus on software, models, and digital technology, cooperating openly with all chip vendors.
At this conference, Tencent Cloud Intelligent Computing made a major upgrade towards "AI Infra closer to Agents", establishing a complete solution that includes Agent Infra solutions + Cloud Mate cloud expert service agents and full-chain security capabilities, helping Agentic AI move from "laboratory" to "production grade".
Li Li, Vice President of Tencent Cloud, said that as the number of Agents increases explosively, enterprise demand for cloud computing infrastructure will quickly upgrade from "resource supply" to "business value". AI Infra must provide faster inference efficiency, more flexible tool integration, more reliable system assurance, and more automated service capabilities.
Reportedly, the core advantage of Tencent Cloud Intelligent Computing lies in "same-source isomorphism"—no need to build AI-specific storage and other facilities; relying on existing cloud-native capabilities refined over the years, it can meet intelligent computing needs and achieve deep integration of cloud and intelligent computing capabilities.
AI Landscape
Tencent Cloud is Tencent's window for exporting AI capability. At this Global Digital Ecosystem Summit, Tencent Cloud also announced it would fully open its accumulated technology and AI practices, connecting both C-end and B-end technologies and scenarios.
For Tencent Cloud, this is undoubtedly a milestone moment. Last year, Tencent Cloud clearly identified AI, overseas expansion, and integrated innovation as its primary future growth channels. At this conference, AI and overseas became the two core themes, with more results on display.
Regarding the much-anticipated AI, Tencent has currently built a full-link product matrix around large models—including basic infrastructure, self-developed large models, model development platforms, agent development platforms, and various scenario-oriented intelligent applications—to help enterprise clients rapidly implement large models in their scenarios.
In this conference, Tencent also released for the first time the "Tencent Cloud Agent Strategy Panorama", as well as the Tencent Cloud Agent Development Platform ADP3.0, Agent Infra solution Agent Runtime, Tencent Cloud Expert Service Agent Cloud Mate, and the Hunyuan 3D 3.0 base model.
Notably, at the start of this year, Tencent brought a number of C-end AI products and applications such as Yuanbao, QQ Browser, ima, etc. into Tencent Cloud, which now undertakes the task of developing both B-end and C-end AI products.
As the pioneer of Tencent's C-end native AI applications, since the beginning of this year, Yuanbao has rapidly upgraded its capabilities around search, document understanding, image processing, etc. At this conference, Yuanbao took the lead together with Tang Daosheng, demonstrating its Q&A capabilities on site.
According to data provided by Tencent, now the amount of user questions to Tencent Yuanbao each day is already equal to the total for a whole month at the beginning of this year, and Yuanbao has become one of the top three AI native apps in China by DAU.
Moreover, Yuanbao continues to deeply integrate with Tencent's existing applications, and is now connected to Tencent Meeting, Tencent Docs, as well as WeChat, Tencent Video, QQ Music, and more than 10 other apps.
Tang Daosheng noted that after the organizational transformation at the beginning of this year, Tencent Cloud has a bigger stage to shine, with more capability to connect C-end and B-end, making both sides’ products more competitive. "So, we are still very excited."
However, Tang Daosheng also told Wallstreetcn that To C’s native AI applications are still in the investment phase, and Tencent Cloud is not yet deeply considering commercial aspects. The current focus remains on refining products and service experience. For example, Yuanbao and ima will next focus more on improving user experience and extending scenario coverage so users can solve real problems more naturally and efficiently.
With continued investment, Tencent's AI application landscape is rapidly unfolding from B-end to C-end. Meanwhile, AI has become Tencent's "new business gene," driving strong growth in its core businesses such as gaming and advertising.
Capital markets have also warmly responded to Tencent's AI strategy achievements, pushing its market value back above 6 trillion HKD on September 17, reaching a near four-year high.
Global tech giants have all joined the drive to reshape businesses with AI, and the "AI quotient" has become a key metric for capital markets to revalue tech firms. Although AI commercialization is still in its early stages, Tencent has clearly secured a leading position.
Risk Warning and DisclaimerThe market involves risk; investment requires caution. This article does not constitute individual investment advice and does not take into account the specific investment objectives, financial situations, or needs of individual users. Users should consider whether any opinions, viewpoints, or conclusions contained in this article are applicable to their particular circumstances. Investing based on this is at your own risk. ```