Tencent Announces Full Support for Mainstream Domestic Chips; Cloud Overseas Customer Base Doubles in One Year

Key Points

Platform upgrades: ADP 3.0 and Agent Runtime enhance agent development and infra; the cloud sandbox can start in 100 ms and support hundreds of thousands concurrent Agent instances.
Cloud Mate impact: Cloud Mate has delivered over one million services, achieves a 95% interception rate for risky SQL, and cut troubleshooting from 30 hours to 3 minutes in real environments.
Hunyuan 3D leap: Hunyuan 混元 released Hunyuan 3D 3.0, improving modeling precision roughly threefold with a geometric resolution of 1536³, and offers a free Hunyuan 3D AI creation engine.
Compute & chip strategy: Tencent now fully supports mainstream domestic chips (国产芯片), pursuing full‑stack software/hardware co‑optimization and heterogeneous compute to lower AI compute costs.
Global growth: International customers have doubled in the past year across > 80 countries and regions, with plans spanning 21 regions and over 3,200 global acceleration nodes.

Tencent Cloud Agent Runtime Capabilities
Capability	Description
Execution Engine	Manages the execution of agent tasks and workflows.
Cloud Sandbox	Provides an isolated, secure environment for agent operation; starts in 100ms, supports hundreds of thousands of concurrent instances.
Gateway	Handles ingress and egress traffic for agents, ensuring secure and efficient communication.
Context Management	Maintains and manages the state and context for long-running agent interactions.
Observability	Provides tools and metrics for monitoring agent performance, logs, and diagnostics.

Key Enhancements and Impact of Cloud Mate
Feature/Metric	Description/Value
Services Delivered	Over one million intelligent architecture-governance services.
Risk Interception Rate	95% for risky SQL in internal practice.
Troubleshooting Efficiency Improvement	Reduced from 30 hours to as fast as 3 minutes.
Primary Function	Visualized cloud architecture governance.
Benefits	Front-loaded risk interception, dramatically faster troubleshooting.

Tencent’s Hunyuan 3D 3.0 Model Capabilities
Capability	Description/Feature
Model Version	Hunyuan 3D 3.0
Key Innovation	Novel 3D-DiT hierarchical sculpting method
Precision Improvement	Roughly threefold versus the prior generation.
Geometric Resolution	1536³
Availability	Integrated into free Hunyuan 3D AI creation engine; accessible via Tencent Cloud APIs.
Professional Tool	Hunyuan 3D Studio (invite-only internal beta) for comprehensive 3D creation pipeline.

Key Aspects of Tencent’s Open-Source and Model Ecosystem Strategy
Aspect	Details
Models Released (YTD)	More than 30 models from the Hunyuan family.
Open-Source Status	Progressively open-sourced.
Included Releases	Hybrid inference models (Hunyuan-A13B), multilingual translation models (30+ languages), full multimodal generation toolchains (image, video, 3D).
Open-Source Focus 1	Delivering real, battle-tested use cases from production scenarios.
Open-Source Focus 2	Providing best-practice playbooks with clear problem-resolution steps for developers.
Strategic Importance	Emphasized as a key industry topic in the large-model era by Tencent Cloud VP, Wu Yunsheng.

Tencent Cloud’s International Growth Metrics and Infrastructure Plans
Metric/Plan	Details
International Business Growth (3 yrs)	Maintained high double-digit growth.
Internet Companies Adoption	Cloud provider for >90% of internet companies expanding globally.
Game Companies Adoption	Cloud provider for >95% of leading global game companies.
Q2 2025 International Revenue	Grew significantly year-over-year.
Overseas Customers Growth (past year)	Doubled, covering over 80 countries and regions.
New Data Centers Planned	Saudi Arabia and Osaka.
Global Infrastructure Span	21 regions.
Global Acceleration Nodes	More than 3,200.

Tencent Cloud domestic chip support became a headline at the Tencent Global Digital Ecology Conference on September 16, 2025.

Overview — Dual Growth Engines: Intelligence and Globalization

On September 16, 2025, the Tencent Global Digital Ecology Conference was held in Shenzhen.

The event showcased a series of AI technology and product updates and announced that Tencent (Tengxun 腾讯) will fully open its AI deployment capabilities and leading industry scenarios.

Tang Daosheng (Tang Daosheng 汤道生), Tencent Group Senior Executive Vice President and CEO of Cloud & Smart Industries Group (CSIG), framed the strategy as two core growth drivers.

He said that “extracting efficiency through intelligence and expanding revenue through globalization” have become the two core drivers of corporate growth.

“Tencent will build two efficiency engines: ‘Intelligentization’ and ‘Globalization,’” he emphasized.

Resume Captain

Your AI Career Toolkit:

AI Resume Optimization
Custom Cover Letters
LinkedIn Profile Boost
Interview Question Prep
Salary Negotiation Agent

Get Started Free

Intelligence: Major AI Platform and Model Upgrades

Tencent Cloud (Tengxun Yun 腾讯云) publicly released a comprehensive intelligence strategy.

Tencent announced broad access to its AI capabilities across both consumer (C‑end) and enterprise (B‑end) scenarios.

Platform and developer tools

Tencent Cloud’s Intelligent Agent Development Platform (ADP) released version 3.0 globally, continuing iterative support for LLM+RAG, Workflow, Multi‑Agent frameworks and other agent development paradigms.
Tencent introduced a multimodal intelligent data lake called TCLake to build a unified data foundation and accelerate Agent application deployment.

Agent infrastructure

On the infrastructure side, Tencent Cloud launched the Agent Infra solution “Agent Runtime”.

Agent Runtime provides five core capabilities: execution engine, cloud sandbox, gateway, context management and observability.

The cloud sandbox — a core component of Agent Runtime — can start in as little as 100 ms and supports concurrent operation of hundreds of thousands of Agent instances.

Expert services agent — Cloud Mate

Cloud Mate is an expert services agent unveiled by Tencent Cloud to help with visualized cloud architecture governance.

Cloud Mate offers front‑loaded risk interception and dramatically faster troubleshooting.

According to data Tencent shared with Kechuangban Ribao, Cloud Mate has delivered over one million intelligent architecture‑governance services to customers.

In internal practice, Cloud Mate achieves a 95% interception rate for risky SQL.

Troubleshooting efficiency has fallen from an average of 30 hours to as fast as 3 minutes with Cloud Mate in real environments.

Hunyuan model upgrades and 3D capabilities

Tang Daosheng announced a new upgrade to Tencent’s Hunyuan (Hunyuan 混元) large model family.

The Hunyuan 3D 3.0 model was released at the conference.

It introduces a novel 3D‑DiT hierarchical sculpting method, improving modeling precision roughly threefold versus the prior generation.

Hunyuan 3D achieves a geometric resolution of 1536³.

Hunyuan 3D is integrated into the Hunyuan 3D AI creation engine, offered free to users, and accessible via Tencent Cloud APIs.

For professional art workflows, Hunyuan 3D Studio has launched an invite‑only internal beta that rebuilds the 3D creation pipeline across seven core steps, including component generation and UV unwrapping.

Open‑source and model ecosystem

So far this year, the Hunyuan family has released more than 30 models and has been progressively open‑sourced.

Releases include hybrid inference models (Hunyuan‑A13B), multilingual translation models supporting 30+ languages, and full multimodal generation toolchains for image, video and 3D.

Wu Yunsheng (Wu Yunsheng 吴运声), Tencent Cloud Vice President, has emphasized open source as a key industry topic in the large‑model era.

Tencent’s open‑source approach focuses on delivering real, battle‑tested use cases derived from production scenarios.

It also focuses on providing best‑practice playbooks that include clear problem‑resolution steps so developers can reproduce solutions in new contexts.

Compute and domestic chip support

Addressing industry concerns around compute, Qiu Yuepeng (Qiu Yuepeng 邱跃鹏), Tencent Group Vice President and President of Tencent Cloud, announced that Tencent has fully adapted its stack to mainstream domestic chips (guochan chips 国产芯片).

Tencent is also actively contributing to and participating in open‑source communities.

Tencent Cloud’s long‑term strategy includes full‑stack software/hardware co‑optimization.

Through a heterogeneous compute platform and associated software capabilities, Tencent aims to integrate different chip types to offer cost‑effective AI compute to customers.

Find Top Talent on China's Leading Networks

Post Across China's Job Sites from $299 / role, or
Hire Our Recruiting Pros from $799 / role

- - - - - - - -

Qualified Candidate Bundles
Lower Hiring Costs by 80%+
Expert Team Since 2014

Get 25% Off
Your First Job Post

Globalization: Accelerating international growth

Tencent Cloud presented its latest globalization results and international product upgrades.

Over the past three years, Tencent Cloud’s international business has maintained high double‑digit growth.
Tencent Cloud is the cloud provider of choice for more than 90% of internet companies and over 95% of leading game companies that go global.
A range of Chinese companies — including Meituan (Meituan 美团), Midea (Midea 美的), Pop Mart (Paopaomate 泡泡玛特) and Yang Guofu (Yang Guofu 杨国福) — have relied on Tencent Cloud to accelerate their overseas expansion.
In 2025 Q2, Tencent Cloud’s international revenue grew significantly year‑over‑year, and the number of overseas customers doubled in the past year, covering more than 80 countries and regions.
Tencent Cloud plans to add new data centers in Saudi Arabia and Osaka, and continues to build a global infrastructure that will span 21 regions and more than 3,200 global acceleration nodes.

International product localization

Tencent Cloud has completed multiple international upgrades so its products better fit global technical ecosystems.

Key products — including the ADP agent development platform, the AI full‑stack engineer tool CodeBuddy, and the enterprise full‑channel commerce solution Cloud Mall — now have international versions.

These products also include deeper integration with large‑model AI technologies for better localization and developer adoption.

What this means

Tencent’s announcements signal a two‑pronged strategy.

First, broaden AI adoption through platforms, infrastructure and real‑world, open‑source use cases.

Second, expand Tencent Cloud’s global footprint with localized products and accelerated infrastructure.

Full support for mainstream domestic chips and continued open‑source contributions may accelerate the domestic AI compute ecosystem and diversify customers’ hardware choices.

For investors, founders, techies, and marketers, the practical takeaways include:

Faster Dev Cycles: Agent Runtime and ADP 3.0 reduce iteration time for LLM‑driven agents and workflows.
Operational Risk Reduction: Cloud Mate’s high interception rate and reduced troubleshooting times lower operational costs and incident windows.
3D Content Acceleration: Hunyuan 3D 3.0 and the Hunyuan 3D Studio beta aim to accelerate professional 3D pipelines and lower entry barriers for creators.
Compute Flexibility: Full support for mainstream domestic chips unlocks alternative procurement strategies and potential cost optimization for AI workloads.
Global Expansion Playbook: Tencent Cloud’s international growth, customer wins, and infrastructure plans provide a replicable path for Chinese companies scaling overseas.

What to watch next:

How open‑sourced Hunyuan models perform in cross‑border developer communities.
Adoption rates for ADP 3.0 and Agent Runtime among global ISVs and system integrators.
Real adoption metrics for domestic chip support in production AI workloads.