Malaysian businesses experimenting with AI just gained a bigger local backbone to build on, as Alibaba folds a new Malaysia data centre into a wide push to scale its AI operations.
Editor
Kai T chevron_right
Table of Contents
Alibaba scales its AI infrastructure
Alibaba has laid out a set of milestones from the first half of 2026 that it says reflect steady progress on its AI strategy. The piece most relevant to Malaysia sits in the infrastructure column: as part of a US$53 billion AI infrastructure commitment, the company added new data centres in Japan, Malaysia, France and Mexico. That brings Alibaba Cloud to 105 availability zones across 32 regions, which in practice means lower-latency cloud and AI services for customers running workloads closer to home.
Alibaba is framing the effort around a full stack that spans its own chips, cloud infrastructure, foundation models and applications, the layers it argues are needed to compete as companies move from AI experiments to wider adoption.
One home for its AI teams
In March, Alibaba created the Alibaba Token Hub business group under chief executive Eddie Wu, pulling its Tongyi laboratory, its model-as-a-service line, and its Qwen, Wukong and AI innovation units under one roof. The stated mission is narrow on purpose: to create, deliver and apply tokens, the basic unit of work for large AI models.
The model and product releases over the half year include:
- Qwen3.7-Max (May), a large language model built for agentic coding, complex reasoning and long-running tasks.
- HappyHorse 1.1 (June), a video generation model aimed at short-form content, advertising and game cinematics.
- HappyOyster 1.0 (June), an interactive world model for films, games and visual experiences.
- The Qwen App, a consumer assistant that ties Taobao, Alipay, Fliggy and Amap into one conversational interface.
- Qwen Glasses, shown at MWC Barcelona, offering real-time translation, capture and payments.
- The Wukong Platform (March), an enterprise agentic system for multi-step workflows.
Why it matters
For Malaysian developers and enterprises, the headline is local capacity. A nearby data centre region tends to translate into faster, more reliable access to cloud and AI tools, and it deepens the options beyond the US hyperscalers that have dominated the market here. Alibaba is also pointing to applied work through its DAMO Academy, including AI screening tools for fatty liver disease and colorectal cancer, as evidence that the models are moving past demos.
The takeaway
Much of Alibaba’s announcement is corporate housekeeping aimed at investors, but the Malaysia data centre is the part that lands closest to home. More regional capacity means local teams can run AI and cloud workloads with less distance between them and the servers doing the work.
Images courtesy of Taylor Vick and Brecht Corbeel on Unsplash.