# Alibaba Cloud Charts Ambitious Global Expansion as AI Demand Surges
Alibaba has unveiled plans to scale its global data centre footprint to over 20 gigawatts by the end of the decade, signaling a dramatic increase in infrastructure spending driven by the rapidly growing demand for AI computing power. The announcement came from Alibaba Group CEO Eddie Wu during a major technology conference in Hangzhou, where he outlined a strategy that extends far beyond simply building more facilities.
## A Massive Financial Commitment Behind the Growth
The 2032 target follows a sweeping financial pledge made earlier this year. In February 2025, Alibaba committed to pouring over 380 billion yuan — roughly $52 billion at the exchange rates of that period — into cloud computing and AI infrastructure across a three-year period. The company stated that this single investment commitment would surpass what it had spent on those same areas over the preceding ten years combined.
Reports indicate that Alibaba has already deployed approximately half of this planned funding during 2026 alone, accelerating spending on proprietary chips and infrastructure to meet customer demand.
## Building an End-to-End AI Technology Stack
Alibaba’s expansion strategy is not limited to physical data centre capacity. The company is pursuing development across every layer of the AI technology stack:
– **Custom AI accelerators** designed for training and inference workloads
– **Server processors** for general-purpose and AI-specific computing
– **Networking and storage systems** tailored for large-scale AI clusters
– **Cloud services** to deploy and manage AI applications
– **Large language models**, including the Qwen family
Wu described this as a comprehensive, vertically integrated approach that covers models, semiconductors, and data centres in unison.
## Strong Financial Momentum in AI Cloud Services
The financial results underscore why Alibaba is investing so aggressively. In the June quarter of 2026, revenue from AI Cloud and Compute Services hit US$7.1 billion, representing 45% year-over-year growth. Adjusted earnings before interest, taxes, and amortisation for the cloud division more than doubled, reaching US$830 million, while the segment’s adjusted EBITA margin climbed to approximately 12%.
Notably, Alibaba’s AI-related product revenue has recorded triple-digit growth for twelve consecutive quarters, reflecting sustained and accelerating demand from enterprise customers.
## Custom Silicon: The Zhenwu V900 and Beyond
A key element of Alibaba’s strategy is reducing reliance on external chip suppliers. The company introduced the Zhenwu V900, an AI processor built by its T-Head semiconductor division. The V900 features 216GB of memory and 1,200GB/s of inter-chip bandwidth, with native support for efficient data formats such as FP8 and FP4.
Alibaba claims the V900 delivers triple the performance of its predecessor, the Zhenwu M890, which debuted in May. Mass production of the V900 is expected to begin in the first quarter of 2027.
Beyond accelerators, Alibaba has also announced plans for its Yitian 720 and Yitian 730 general-purpose CPUs in 2027, with the 730 based on T-Head’s own microarchitecture. The company has already shipped over 560,000 Zhenwu processors to more than 400 external customers spanning 20 different industries.
## Advanced Networking and Storage for AI Clusters
As AI compute clusters grow in size and complexity, the interconnects and storage systems that tie them together become critical bottlenecks. Alibaba has responded with significant upgrades to its networking and storage infrastructure:
– **HPN 8.0 Pro networking architecture**, offering 100 petabits of bandwidth and support for over 130,000 network ports operating at 800Gbps within a single cluster
– **Cloud Parallel File Storage**, engineered for AI training with throughput in the hundreds of terabytes per second and hundreds of millions of IOPS, with Alibaba claiming a 69% reduction in AI storage costs
The company is also building a new-generation supernode server that integrates the V900 accelerator with custom networking, storage, and interface components, designed to support clusters of up to 500,000 accelerator cards.
## Supply-Demand Imbalance and Scaling Challenges
Despite the ambitious expansion, Alibaba acknowledged that customer demand for AI services is currently outpacing its available supply. Wu noted that supply-chain constraints are limiting the pace at which the company can bring new capacity online. The company plans to begin deploying AI supernodes at commercial scale during the current quarter to address this gap.
Wu also highlighted that margins are expected to improve over time as Alibaba increases the use of its own T-Head processors, gradually replacing commercially purchased chips in its data centres with domestically designed alternatives.
## A Broader Industry Trend
Alibaba is far from alone in pursuing custom infrastructure. Major hyperscalers around the world are following a similar path:
– **AWS** offers Trainium AI processors and Graviton CPUs, with over 90,000 customers now using Graviton-based infrastructure
– **Microsoft** has deployed its Maia accelerators and Cobalt CPUs across Azure, with Cobalt now present in nearly half of its data centre regions
– **Google** combines its own TPUs and Axion CPUs with third-party Nvidia, Intel, and AMD hardware in its AI Hypercomputer architecture
All three providers continue to use third-party hardware alongside their own silicon, reflecting the reality that a fully self-contained supply chain remains a long-term goal rather than an immediate replacement. Amazon has pointed to improved price-performance as a key driver, reporting that Trainium2 offers roughly 30% better price-performance than comparable GPUs, with Trainium3 adding another 30% to 40% on top of that.
## What This Means for the Global AI Landscape
Alibaba’s expansion signals that the race to build AI infrastructure is intensifying on a global scale. With a 20-gigawatt target, investments exceeding the previous decade’s spending in a single three-year window, and a vertically integrated approach spanning hardware, networking, storage, and software, Alibaba is positioning itself to capture a growing share of the AI cloud market — particularly in the Asia-Pacific region and among Chinese enterprise customers seeking domestic technology solutions amid ongoing global trade restrictions on advanced semiconductors.
—
## Frequently Asked Questions
**Why is Alibaba investing so heavily in data centre capacity?**
Alibaba is responding to surging demand for AI computing services. The company’s AI Cloud revenue grew 45% year over year in the latest quarter, and customer demand has outstripped available supply. The 20-gigawatt target by 2032 reflects the scale of infrastructure needed to meet this demand.
**What is the Zhenwu V900 processor?**
The V900 is Alibaba’s latest AI accelerator, developed by its T-Head semiconductor unit. It features 216GB of memory, 1,200GB/s of inter-chip bandwidth, and support for efficient data formats like FP8 and FP4. It is three times faster than the previous M890 model and is expected to enter mass production in early 2027.
**How does Alibaba’s investment compare to other cloud providers?**
Alibaba’s three-year commitment of over 380 billion yuan exceeds its cumulative investment in cloud and AI infrastructure over the previous ten years. While direct comparisons with other hyperscalers are difficult due to differences in scale and structure, the commitment places Alibaba among the most aggressive investors in AI infrastructure globally.
**Why is Alibaba developing its own chips?**
Custom silicon gives Alibaba greater control over its supply chain, improves cost efficiency, and reduces dependence on external suppliers — particularly important given US restrictions on advanced AI chip exports to China. Over time, using proprietary processors is also expected to improve the company’s margin profile.
**What are the main bottlenecks Alibaba is facing?**
Supply-chain constraints are currently limiting how quickly Alibaba can expand its data centre capacity. Despite strong demand and significant investment, the physical build-out of infrastructure takes time, and the company has acknowledged that demand is exceeding current supply.
—
## Conclusion
Alibaba’s announcement represents one of the most ambitious infrastructure expansion plans in the cloud computing industry to date. By targeting over 20 gigawatts of data centre capacity by 2032, investing more in three years than it did in the previous decade, and building a vertically integrated AI technology stack that spans semiconductors, networking, storage, and cloud services, Alibaba is making a decisive bet on the long-term growth of AI-driven computing demand.
The combination of strong financial results — with 45% year-over-year revenue growth in AI cloud services and twelve consecutive quarters of triple-digit AI product revenue growth — suggests that the demand is real and sustained. However, supply constraints and the sheer scale of the build-out present significant execution challenges that will shape Alibaba’s trajectory in the years ahead.
As the global race to build AI infrastructure continues to accelerate, Alibaba’s moves serve as a clear reminder that the companies best positioned to capture the AI opportunity will be those that invest early, invest boldly, and build across the full technology stack.
Thank you for reading
