Following the technological trajectory of Google, Alibaba is building its own full-stack AI high ground under the banner of &#34;Tongyun Ge&#34; (Tongyi Lab + Alibaba Cloud + T-Head).
Shenwan Hongyuan's latest research report points out that among global leading cloud providers, only Google and Alibaba have chosen a &#34;full-stack in-house R&amp;D&#34; path, pursuing top-tier self-research capabilities simultaneously across cloud computing, foundational large models, and AI chips.
Core data currently validates the strategy's effectiveness: in the first half of 2025, Alibaba Cloud holds a 35.8% share in the Chinese AI cloud market, ranking first in the industry and exceeding the combined share of the 2nd to 4th players; the self-developed &#34;Zhenwu 810E&#34; chip has achieved deployment in clusters of tens of thousands of cards; and within 9 hours of the launch of the Tongyi Qianwen &#34;3 Billion Free Orders&#34; event, AI orders surpassed 10 million.
For investors, Alibaba is at a critical window for AI commercialization, transitioning from an &#34;investment phase&#34; to a &#34;monetization phase.&#34; Triple catalysts—cloud business price adjustments, the establishment of the Token Hub business group, and the explosion in demand for computing power—are initiating simultaneously. Coupled with a five-year compound annual growth rate of 26.8% for the industry, the current stock price possesses a high margin of safety.
<h2>Google: The Global Benchmark for a Full-Stack AI Closed-Loop, Providing a Reference for Alibaba's Valuation</h2>
Google has established a complete closed-loop of &#34;computing power—cloud platform—large model—application,&#34; providing a direct valuation benchmark for understanding Alibaba's &#34;Tongyun Ge&#34; strategy.
Computing Power Layer: The seventh-generation TPU, Ironwood, officially opened in November 2025, offering over 4 times the training and inference performance per chip compared to the previous generation TPU v6 Trillium. It features a network bandwidth of 9.6Tb/s, supports high-speed interconnection of 9,216 chips, and can access 1.77PB of shared high-bandwidth memory (HBM).
Cloud Platform Layer: Google Cloud's 2025 full-year operating revenue exceeded $60 billion. Quarterly revenue in Q4 reached $17.664 billion, a year-on-year increase of 48%; quarterly operating profit reached $5.313 billion, a year-on-year surge of 154%, with profitability release accelerating.
Large Model Layer: Gemini 3 has formed a full-size matrix of Ultra/Pro/Flash/Nano. By the end of 2025, it was processing over 100 billion tokens per minute, and monthly active users surpassed 750 million. The enterprise product, Gemini Enterprise, achieved over 8 million paid seats within just four months of its launch.
<h2>Alibaba Cloud: China's #1 AI Cloud, Signaling a Profitability Turning Point with Price Adjustments</h2>
Alibaba Cloud's leading position in terms of market share is quite solid. According to Omdia data, the Chinese AI cloud market scale reached 22.3 billion yuan in the first half of 2025, with Alibaba Cloud ranking first with a 35.8% share, exceeding the combined total of the 2nd to 4th players. In the public cloud IaaS market, it also leads with a 21.31% share. In the third quarter of 2025, the total scale of the cloud infrastructure service market in mainland China was $13.4 billion; Alibaba Cloud accounted for 36%, while Huawei Cloud and Tencent Cloud held 16% and 9% respectively, showing a significant market concentration effect.
Performance growth is accelerating. In the fourth quarter of 2025, Alibaba Cloud's revenue increased by 36% year-on-year to 43.3 billion yuan, and its adjusted EBITA margin improved to 9.04%. At the capital expenditure level, Alibaba's Capex for the quarter reached 29 billion yuan, significantly higher than Tencent's 19.6 billion yuan and Baidu's 2 billion yuan during the same period, reflecting its sustained heavy investment in AI computing power infrastructure.
Price adjustments signal positive momentum. Starting April 18, 2026, Alibaba Cloud will adjust prices for its AI computing power and storage products. Prices for computing cards like Zhenwu 810E will increase by approximately 5%–34%, while CPFS (Intelligent Computing Edition) will rise by about 30%. This move aligns with the trend of Amazon AWS and Google Cloud increasing prices for their core AI products in early 2026, potentially marking a new pricing cycle for the industry and directly benefiting the improvement of Alibaba Cloud's profitability.
Long-term growth paths remain clear. Omdia predicts that the compound annual growth rate of the Chinese AI cloud market from 2025 to 2030 will reach 26.8%, driven primarily by structural trends such as multi-agent collaboration, context engineering optimization, increased demand for AI security, and large-scale API calls.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/c5bce531-1215-4518-8053-dffc24c25a83.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="Image" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/c5bce531-1215-4518-8053-dffc24c25a83.png"/>
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/fc45bc05-031c-4f7a-ad39-3f2450526f42.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="Image" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/fc45bc05-031c-4f7a-ad39-3f2450526f42.png"/>
<h2>Tongyi Qianwen: The World's Largest Open-Source Model Ecosystem, Accelerating Validation of the Commercial Closed-Loop</h2>
In terms of the open-source ecosystem, Alibaba's Tongyi has established a significant global lead. As of the report's release, Tongyi has cumulatively open-sourced over 300 models, with global downloads exceeding 600 million and the number of derivative models reaching 170,000, all ranking first in the world. Simultaneously, over 1 million enterprise customers have accessed Tongyi's large models, and the ecosystem's coverage continues to expand.
Penetration into the enterprise market is accelerating. According to Frost &amp; Sullivan data, the daily average call volume of enterprise-level large models in China in the second half of 2025 significantly increased by 263% compared to the first half, reaching 37.0 trillion tokens. Among these, Tongyi Qianwen's market share jumped from 17.7% in the first half to 32.1%, an increase of 14.4 percentage points, making it the most chosen large model in the B-end market.
C-end commercialization has achieved a historic breakthrough. On February 6, 2026, Qianwen launched the &#34;3 Billion Free Orders&#34; event, during which AI-generated orders exceeded 10 million within 9 hours. By February 23, nearly 200 million AI orders had been completed, with users issuing 5 billion &#34;Qianwen, help me&#34; commands. On February 7, Qianwen's DAU reached 73.52 million, ranking first on the Apple App Store free chart for 8 consecutive days. Notably, over 4 million users over the age of 60 used Qianwen to experience AI food ordering for the first time, fully validating the cross-demographic penetration capability of AI applications.
Model capabilities are also continuously evolving. Starting March 31, 2026, Tongyi released three major model updates over three consecutive days. Among them, Qwen3.5-Omni achieved SOTA (State-Of-The-Art) results in 215 tasks, with some core capabilities surpassing Gemini-3.1 Pro, showcasing the strength to compete on the same stage as top international models.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/0782c8e9-fbbc-4fd9-b203-4bfb707ac1af.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="Image" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/0782c8e9-fbbc-4fd9-b203-4bfb707ac1af.png"/>
<h2>T-Head Zhenwu 810E: Self-Developed Chip Achieves Tens of Thousands of Cards Commercial Validation</h2>
In terms of core parameters, Zhenwu 810E is already capable of competing with high-end international products. The chip is equipped with 96GB of HBM2e high-bandwidth memory, has an inter-chip interconnect bandwidth of 700GB/s, supports PCIe 5.0×16 interface, and maintains power consumption below 400W. Its video memory capacity is on par with the NVIDIA H20 (96GB) and exceeds the NVIDIA A800 (80GB) and Huawei Ascend 910B (64GB), demonstrating competitiveness in key metrics.
The scale of commercial implementation has also exceeded expectations. Zhenwu 810E has been deployed in multiple clusters of tens of thousands of cards within Alibaba Cloud and serves over 400 customers, including State Grid, the Chinese Academy of Sciences, XPeng Motors, and Sina Weibo. This progress signifies that domestic AI chips are accelerating from the laboratory stage into large-scale commercial applications.
The &#34;Chip-Cloud Integration&#34; architecture has built a differentiated competitive barrier for Alibaba Cloud. Chip computing power is deeply coordinated with Alibaba Cloud's complete AI framework, platform, models, and applications, forming an integrated product and service system. Compared to third-party procurement models, this synergy significantly enhances computing power scheduling and optimization efficiency, and also builds a moat of cost and performance that is difficult to replicate.
In addition, the release of XuanTie C950 further expands Alibaba's layout in the RISC-V ecosystem. Launched in March 2026, this chip scored over 22 points/GHz in the SPECint2006 single-core general performance test, a more than 3x improvement in overall performance compared to the previous generation, and natively supports 100-billion-parameter large models such as Qwen3 and DeepSeek V3, solidifying Alibaba's technical foundation in the open computing ecosystem.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/14900662-8119-4ab6-a68a-93db5ffed7df.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="Image" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/14900662-8119-4ab6-a68a-93db5ffed7df.png"/>
<h2>Qianwen Personnel Changes: Short-Term Disturbance, Strategic Direction Unchanged</h2>
In early March 2026, the Tongyi Qianwen team experienced personnel changes: Lin Junyang, the head, and Yu Bowen, the head of post-training, both resigned on the same day. Previously, Hui Binyuan, the head of Qwen Code, had already left in January to join Meta. Analysts believe this change has a neutral impact, based on the following core logic:
Clear Strategic Determination from Management. CEO Eddie Wu emphasized in an internal email that developing foundational large models is a critical strategy for Alibaba's future, and investment in AI R&amp;D and the recruitment of outstanding talent will continue to increase.
Rapid Replacement by Top External Talent. Zhou Hao, a former Senior Staff Research Scientist at DeepMind (who was involved in key projects like Gemini 3.0), joined in January 2026 to take over as the head of the post-training team.
Structural Upgrade of the R&amp;D System. This adjustment is seen as a proactive transformation from a single-team closed-loop model to a multi-module parallel and group-level coordinated system, rather than a contraction of capabilities. Zhou Jingren, CTO of Alibaba Cloud and head of Tongyi Lab, will continue to lead subsequent work.
Strengthening the Layout of the Token Hub Strategy. On March 16, Alibaba officially established the Alibaba Token Hub (ATH) business group, personally overseen by CEO Eddie Wu, to build a unified scheduling center focused on &#34;creating Tokens, delivering Tokens, and applying Tokens,&#34; aiming to position itself for high-value competition in the AI Agent era.
<h2>From Valuation Models to Core Beneficiary of AI Infrastructure</h2>
The report maintains its forecasts for Alibaba's Non-GAAP net profit attributable to shareholders for fiscal years 2026-2028 at RMB 79.0 billion, RMB 109.5 billion, and RMB 149.2 billion, respectively.
Based on the SOTP valuation system, the valuation breakdown for each business segment is as follows:
<ul>
<li>
Taobao and Tmall Group: Given a 10x PE for FY26E, corresponding to a valuation of RMB 1.9427 trillion;
</li>
<li>
Alibaba Cloud: Given a 10x PS for FY26E, corresponding to a valuation of RMB 1.5868 trillion;
</li>
<li>
International Digital Commerce Group: Given a 2x PS for FY26E, corresponding to a valuation of RMB 290.8 billion;
</li>
<li>
Other Businesses: SOTP valuation of RMB 57.4 billion.
</li>
</ul>
The group's overall valuation reaches RMB 3.8777 trillion, corresponding to a target price of $234 per share, representing a 96% upside potential from the current $119.72.
Alibaba has completed a critical leap from concept validation to commercial implementation in its full-stack AI strategy. Whether it is the large-scale scheduling of cloud computing power, its leading global position in the open-source model ecosystem, or the first large-scale empirical evidence of an AI shopping closed-loop on the consumer end, all point to the same conclusion: Alibaba is becoming a core beneficiary of China's AI infrastructure era.
Risk Disclosure and Disclaimer
Market risks exist, and investment requires caution. This article does not constitute personal investment advice, nor does it consider the specific investment objectives, financial situation, or needs of individual users. Users should consider whether any opinion, view, or conclusion in this article is consistent with their specific circumstances. Investment based on this is at your own risk.

09988

BABA

GOOGL

Alibaba is replicating Google's AI roadmap, pursuing in-house capabilities in cloud computing, foundational large models, and AI chips through its "Tongyun Ge" strategy. A research report from Shenwan Hongyuan indicates that Alibaba Cloud holds a 35.8% share in China's AI cloud market, leading its competitors. Alibaba is at a critical turning point for AI commercialization, with cloud business price adjustments and growing demand for computing power providing a margin of safety for its stock price. Google's full-stack AI closed-loop provides a reference for Alibaba's valuation

- Alibaba is building its AI infrastructure under the name "Tongyun Ge," focusing on integrated self-research in cloud computing and AI technologies.  
- In the first half of 2025, Alibaba Cloud captured 35.8% of China's AI cloud market, leading the industry, while its self-developed "Zhenwu 810E" chips achieved extensive deployment.  
- With a projected annual growth rate of 26.8% for the AI cloud market from 2025 to 2030, Alibaba is poised to benefit significantly as it transitions from investment to monetization phases in AI.

Alibaba, Empowered by "Tongyun Ge," is Replicating Google's AI Roadmap