Alibaba Cloud Plans to More Than Double Global Modular Data Center Capacity This Year
0xBroomberg
Alibaba Cloud said it will more than double modular data-center capacity globally this year, compressing large AI data-center delivery to 100 days — a pace that rewrites infrastructure competition with factory-floor speed.
100 days to deliver — how is that possible?
Conventional large data centers take 6–12 months to deliver in China and 12–18 months in the US. Alibaba Cloud's target: 100 days.
The timeline breaks into three phases: 30 days of factory production and site prep → 50 days of on-site installation → 20 days of commissioning and testing. Servers move in as soon as all five core systems are live.
This means → the speed gain is structural, not just labor. Build in the factory, assemble on-site — serial construction replaced by parallel production.
Modularization from 30% to 90% — what changed?
Alibaba Cloud's CUBE 5.0 architecture raises the modularization rate of five core systems — power, cooling, security, intelligence, and fire safety — from an earlier 30% to 90%.
In plain terms = most of the data center is now built, calibrated, and packed into shipping containers at the factory, then crane-lifted into place on-site like building blocks.
The numbers follow: per-kilowatt construction cost down over 10%, compute density per unit area 5–10× the previous generation, transformer count nearly halved.
How far have efficiency and overseas certifications come?
Air-cooled PUE — a measure of energy efficiency where closer to 1.0 is better — stays at or below 1.15; liquid-cooled PUE at or below 1.10. First-pass product testing hit near-100% pass rates.
CUBE 5.0 has completed pilot deployments in Ulanqab and Zhongwei. It holds CE certification under European EN standards and IEC-standard product approvals for Southeast Asia.
This means → the modular platform is not a domestic-only play. It already carries the credentials to ship overseas.
How fast is AI revenue growing?
AI-related product revenue crossed 30% of external commercial revenue for the first time. Latest-quarter revenue reached RMB 8.97 billion; annualized AI revenue topped RMB 35.8 billion.
CEO Eddie Wu (吴泳铭) said on the earnings call: "AI is driving a full upgrade across Alibaba Cloud. Growth is shifting from traditional compute and storage to models, computing power, and Agent services."
This reflects a structural shift in revenue mix — AI is no longer incremental; it is becoming the main engine.
What needs to be proved next?
Wu expects annualized recurring revenue from AI model and application services — including the Bailian MaaS platform (model-as-a-service, where enterprises call AI models on demand) — to pass RMB 10 billion in the June quarter and RMB 30 billion by year-end, adding that high margins in this segment are becoming visible.
Two checkpoints ahead: ① Can infrastructure capacity actually deliver the more-than-double expansion? ② Can the MaaS platform hit the RMB 30 billion annualized target by year-end?
In plain terms = Alibaba Cloud has drawn a roadmap of "build fast, sell well, profit high." Next earnings season is the exam.
Content is for reference only, not financial advice.