When it comes to global cloud computing, there are three giants — AWS (Amazon), Azure (Microsoft), and Google Cloud (GCP). Their annual revenues are in the range of US$50–120 billion. At first glance, you’d think Alibaba Cloud should learn from AWS. Ultimately, both businesses stem from ecommerce giants, and both are the largest players in their respective home markets.
However, recently we came across this interesting article from a Chinese observer who begs to differ. We have put some excerpts of the author’s viewpoints here.
As the author puts it:
“Alibaba and Amazon are highly similar in business — each is the largest e-commerce platform in their respective markets, and their cloud arms are also the biggest in revenue scale. So why not learn from AWS, the global leader, or Azure, the runner-up — but instead from number three, Google Cloud?”
Azure: highly profitable, but unlearnable
The article notes that Azure’s model is the most profitable among the three:
“In Q2 2025, Azure’s operating margin was about 43% (including some other cloud businesses), compared with AWS’s 33% and Google Cloud’s 21%.”
Why? Because Microsoft’s enterprise software ecosystem gives it enormous pricing power:
“One Fortune 50 enterprise said: ‘We use both Azure and AWS, but since our company relies on Microsoft software like Outlook and Teams, Azure’s pricing is much stronger — other cloud vendors have some room to negotiate.’”
Another partner commented:
“Microsoft’s Power BI helps clients integrate scattered data — from Excel, databases, Azure — into visual dashboards that drive decisions. A lot of Azure’s growth comes from customers switching from Looker and Tableau to Power BI.”
That ecosystem advantage explains Azure’s high profitability. The author points out that Chinese vendors are trying, but failing, to replicate it:
“Tencent pushes Tencent Meeting; Alibaba promotes DingTalk and Quark — all to build a software ecosystem and increase user stickiness. But none can match Microsoft’s comprehensive suite for enterprise clients. It’s not that they don’t want to learn — they simply can’t.”
AWS: a philosophy that’s falling behind
AWS still leads in market share, but its growth has slowed.
“AWS’s growth rate now hovers around 15–20%, significantly lower than Azure’s 35%+ and Google Cloud’s 30%+.”
Its long-standing principle of offering a “scalable infrastructure for customers to build on their own” is no longer aligned with today’s demand:
“Self-building takes too much time — clients now want pre-built solutions that can be deployed quickly.”
Meanwhile, Azure and Google Cloud have moved fast with ready-to-use AI tools and enterprise integrations:
“OpenAI’s partnership with Microsoft Azure captured the early AI opportunity. AWS, without its own large model or tools like Copilot, can’t match that. Nor can it tap into core enterprise data through products like SharePoint, Dynamics, or SQL Server.”
“In the AI era, AWS’s growth has started to lag behind Azure and Google Cloud — so naturally, Alibaba Cloud chooses to learn from the latter two rather than the former.”
Google Cloud: rising with Gemini and TPUs
What makes Google Cloud stand out? The author highlights Gemini, Anthropic, and TPUs as key differentiators:
“A Fortune 500 customer said: ‘We used to mainly use Azure for its seamless integration with OpenAI and Microsoft tools. Later, Google’s Gemini began a price war with OpenAI, so we started using Google Cloud too.’”
“In addition to Gemini, Google is also Anthropic’s second-largest shareholder, offering customers both Claude and Gemini — two of the industry’s top large models.”
Google also benefits from its own Tensor Processing Units (TPUs):
“TPUs are roughly 30% more cost-effective than GPUs in training scenarios. Even Apple uses Google’s TPUs for some compute tasks.”
And the coming V7 TPU chip could extend that advantage:
“The next generation of TPUs, shipping next year, will deliver four times the performance of the V6 — the compute gap may widen further.”
Alibaba Cloud: following Google’s three-pronged playbook
“Alibaba Cloud’s strategy now looks strikingly similar to Google’s. First, it’s expanding compute capacity — its capex on computing power now ties with ByteDance and exceeds Tencent.
Second, it’s investing heavily in large models — Qwen (千问) ranks among China’s top-tier LLMs, alongside ByteDance’s Doubao.
Third, it’s developing its own chips (PingTouGe) and GPU pooling systems — raising GPU utilization from 13.3–33.9% to 48.1%, cutting required H20 GPUs by 82% in trials.”
The bottom line
“AWS’s philosophy now feels outdated. Azure’s success relies on Microsoft’s unmatched software ecosystem — impossible to replicate. Google Cloud’s strengths — compute leadership, model performance, and self-developed chips — are exactly what Alibaba Cloud is pursuing.”
In short, Alibaba Cloud isn’t following the biggest or the richest — it’s following the most forward-looking.
