By Alice Ye in AI — 25 May 2026

China's AI Cloud Wars: How Tencent, Alibaba, and Baidu Are Adapting Their Strategies

What is happening in China's cloud computing market?

China's cloud computing industry is undergoing a fundamental transformation as artificial intelligence becomes the primary competitive battleground. The three major players—Tencent Cloud, Alibaba Cloud, and Baidu Intelligent Cloud—are pursuing distinctly different strategies, each facing unique challenges as the market shifts from traditional infrastructure services to AI-native capabilities.

This shift represents more than incremental evolution. Cloud providers are moving from selling storage and computing resources to selling "tokens"—the basic unit of AI model usage. This changes the economics, competitive advantages, and strategic priorities across the sector.

Why the business model is changing

From infrastructure to intelligence

Traditional cloud computing operated on three layers:

IaaS (Infrastructure as a Service): Renting servers, storage, and network capacity
PaaS (Platform as a Service): Providing development tools and middleware
SaaS (Software as a Service): Delivering ready-to-use applications

In the AI era, the value proposition has shifted. Customers increasingly care less about raw computing resources and more about:

Access to powerful AI models
Token pricing and availability
Integrated AI capabilities across their operations
Custom AI solutions for specific industry needs

This transforms cloud computing from a "wholesale" infrastructure business into a "retail" AI services business, where unit economics depend heavily on the cost per token served.

Tencent Cloud: The profit-first approach

Core strategy: efficiency over growth

Tencent Cloud has prioritized profitability from early on, achieving industry-leading gross margins around 50%—significantly higher than competitors. The company publicly celebrated reaching "full-year scaled profitability" in 2025.

This focus stems from Tencent Cloud's origins: it grew organically from WeChat ecosystem needs, serving internal requirements first. The cloud division has maintained strict financial discipline, with leadership emphasizing "operational efficiency, cost structure, and profit levels."

The trade-offs

This efficiency-first approach has consequences:

Strengths:

Strong financial performance with 22% growth in H2 2025
Access to 130 million+ enterprise users through WeChat ecosystem
Healthy revenue structure with growing enterprise services

Weaknesses:

Market share declining in competitive segments
Limited AI model leadership—Tencent's Yuanbao ranks fourth in AI cloud market share (Alibaba 38%, Volcengine second, Baidu third)
Service quality concerns, highlighted by former executive Wu Hongsheng's public criticism of debt collection practices

The challenge ahead

Tencent Cloud faces a strategic dilemma: its Hunyuan AI model hasn't achieved the influence of competitors like Alibaba's Qwen or ByteDance's Doubao. Without leading AI capabilities, the company must compete on service quality—yet its profit-maximization approach may limit investment in customer experience improvements.

The company's heavy reliance on internal WeChat ecosystem demand creates another vulnerability: future growth increasingly requires expanding beyond Tencent's existing properties.

Alibaba Cloud: The investment-heavy leader

Strategy: Scale through aggressive investment

Alibaba Cloud represents the opposite extreme. As market leader, it has prioritized maintaining dominance through massive capital deployment. CEO Wu Yongming committed over 380 billion yuan ($53 billion) for AI infrastructure investment over three years.

This approach has delivered results:

Market leadership with 38% AI cloud market share (Omdia data)
AI-related revenue exceeding 30% of cloud revenue (8.97 billion yuan in a single quarter)
Strong presence across finance, manufacturing, and expanding international clients

The profitability question

However, this scale-first strategy comes with significant costs:

Capital expenditures reached 126 billion yuan in FY2026, primarily for cloud infrastructure and instant retail. Yet Morgan Stanley estimates Alibaba Cloud's EBITDA margin remains around 8-9%—far below Tencent's 50% gross margin.

The issue: Revenue growth from new AI computing capacity is being offset by depreciation costs, limiting margin expansion despite increasing scale.

Path to commercialization

Wu Yongming has signaled a strategic shift: "Our full-stack AI technology investment has officially crossed the initial cultivation phase and entered a positive commercial returns cycle."

Key levers for improved returns:

Model monetization: Alibaba's Qwen AI model, while not industry-leading, benefits from strong enterprise service capabilities. Converting existing cloud customers into AI solution customers represents major revenue potential.
Chip economics: Alibaba's Pingtouge division has shipped 560,000 Zhenwu M890 chips (3x performance improvement), but deployment scale remains insufficient for meaningful cost reduction. The critical question: Can domestic chips scale enough to replace expensive NVIDIA GPUs in training workloads?
Service stability: Multiple outages (including Alipay/Taobao failures in December, Hong Kong datacenter incidents) have damaged reliability perception at a critical growth phase.

Internal challenges:

The departure of Tongyi Qianwen's key leader Lin Junxiang in March 2026 raised questions about internal strategic alignment. In AI's fast-moving landscape, technical leadership stability directly impacts product development continuity.

Baidu Intelligent Cloud: The validated vision struggling for traction

Early advantages evaporating

Baidu represents perhaps the most complex situation. The company pioneered the "cloud-intelligence integration" strategy years before competitors, combining:

Kunlun custom AI chips
PaddlePaddle deep learning framework
Wenxin large language models
Complete AI stack optimization

This early validation gave Baidu cognitive leadership in AI cloud services. The irony: Baidu proved the market opportunity, but Alibaba and Tencent are capturing the value.

Lost differentiation

Model parity:

Consumer AI: ByteDance's Doubao leads with 226 million MAU; DeepSeek peaks at 187 million; Tencent Yuanbao ranks third. Baidu's Wenxin struggles to compete.
Enterprise solutions: While Wenxin + Kunlun + Baidu's platform maintains strong enterprise recognition, annual revenue (~30 billion yuan) significantly trails Alibaba Cloud's 41.6 billion yuan quarterly revenue.

Full-stack commoditization:
Baidu's "complete self-research" differentiation is eroding:

Huawei's Ascend chips hold ~23% domestic AI chip market share vs. Kunlun's 8%+
Alibaba's "Tongyi-Cloud-Ge" system now replicates Baidu's integrated approach
ByteDance's Volcengine uses aggressive token pricing to drive infrastructure demand

The remaining advantage: Domestic chips

Baidu's strongest card may be Kunlun chips, now valued at ~13 billion yuan after Series D funding. The third-generation P800 has powered China's first fully self-developed 30,000-card cluster.

In an AI era where compute capacity defines cloud economics, owning chip IP provides structural cost advantages—similar to how NVIDIA's GPU leadership drove its market dominance.

The Huawei problem:

However, even this advantage faces challenges. When DeepSeek and other domestic models run on domestic silicon, they predominantly use Huawei's Ascend chips and CANN framework—not Kunlun. Huawei's ecosystem maturity creates a gap Baidu hasn't closed.

Additionally, P800 delivery timelines may face constraints similar to those affecting competitors. Supply chain challenges limiting Huawei likely also impact Kunlun and Alibaba's Pingtouge.

What determines who wins?

The shift to token-based economics creates new competitive dynamics:

Critical success factors:

Model capability: Stronger models drive higher token demand, creating virtuous cycles of usage and infrastructure revenue
Chip economics: Unit token cost depends heavily on computational efficiency. Companies with proprietary, high-performance chips (or preferred access) gain structural advantages
Service ecosystem: Converting infrastructure customers into AI solution customers requires strong integration capabilities and industry expertise
Scale vs. efficiency balance: Pure profit focus (Tencent) risks market share loss; pure scale focus (Alibaba) delays returns; finding the optimal balance becomes crucial

The new competitive equation:

Cloud computing is no longer about selling megabytes or compute hours—it's about selling tokens at competitive unit economics while maintaining model quality. Success requires simultaneously optimizing:

Per-token serving costs (chip efficiency, model optimization)
Token demand generation (model capabilities, application ecosystem)
Customer acquisition and retention (service quality, solution integration)

What happens next?

Each company faces distinct challenges requiring strategy adjustment:

Tencent must decide whether maintaining profit margins justifies market share erosion, or if AI's strategic importance demands increased investment in model capabilities and service quality.

Alibaba enters a commercialization phase after massive investment, needing to demonstrate that scale advantages can translate into sustainable profitability without sacrificing competitive position.

Baidu must leverage its early technical validation into concrete business advantages, likely by maximizing Kunlun chip deployment to create cost structures competitors cannot match.

The underlying question for all three: As AI capabilities become more democratized and chip supply constraints ease, what creates sustainable competitive advantage in token economics? The answer will likely determine not just cloud market rankings, but the broader trajectory of China's AI industry development.

Related Coverage：

Cloud Giants Push Through AI-Driven Price Hikes as Token Demand Strains Compute and Networks