Alibaba Cloud Malaysia has launched a new public cloud region in Johor, bringing two new data centres online and taking its total Malaysia footprint to five facilities. The announcement, made in Kuala Lumpur on 9 June, is paired with plans to roll out a suite of agentic AI services to Malaysian enterprises in the second half of this year, making this more than just an infrastructure expansion.
The Johor expansion forms part of Alibaba’s previously announced US$53 billion investment in AI and cloud infrastructure, and brings its global network to 104 availability zones across 32 regions. The company has operated in Malaysia since 2017, starting with a twin data centre presence before adding a third facility last year.
The new region covers cloud computing products spanning compute, storage, containers, networking, big data, security, databases, and cloud-native services. Being a public cloud region, it serves Malaysian customers as the primary market but remains accessible to businesses in other countries, including Singapore, according to Choong Hon Keat, General Manager of Malaysia at Alibaba Cloud Intelligence.
The agentic AI product suite coming to Malaysia in the second half of this year–AgentRun, STAROps, ACS Agent Sandbox, Agent Security Centre, AI Security Guardrails 2.0, and Agentic SOC–is already generally available globally, confirmed Lin En Shu, head of solutions architecture at Alibaba Cloud Intelligence Malaysia, when asked directly during the media briefing.
But the products are not all the same type of tool, and Lin distinguished that the official materials did not make it clear. AgentRun and STAROps sit in the enterprise infrastructure tier. They are for companies that want to build and run large-scale fleets of AI agents and offer them as a product or service to their own customers. He cited ILMU Claw, a product by YTL AI Labs built on Alibaba Cloud’s infrastructure, as a local example of what that enterprise tier looks like in practice.
The other tier, Qoder and QoderWork, are consumer-facing, interface-driven tools that require no complex engineering to start using.
“AgentRun is really for agentic infrastructure for you to build massive-scale AI agents,” Lin said. “We do have another suite of agentic AI, which is very simple. You just use an interface, you can just talk to it, you can just use it straight away.”
That separation is worth understanding before any enterprise starts evaluating the product list. A company deploying internal AI workflows is looking at different tools than one building agent-powered services for end customers.
Since April this year, Alibaba Cloud has released four versions of its Qwen model family–Qwen3.6-35B, Qwen3.6-27B, Qwen3.7-Max, and Qwen3.7-Plus–a pace that prompted a question from the floor about how enterprises are expected to keep up, and what happens to existing integrations when a newer version arrives.
Lin said most customers simply move to the latest version because newer models are more capable across coding, reasoning, and understanding user intent. But the more useful framing he offered was around model selection, whereby enterprises do not need the most powerful or most expensive model for every task, and should not try to use one that way.
“You don’t need the best model, the most expensive model all the time. As long as the model is good enough for the task, you can go for a cheaper model because you need to use it at scale,” he said. Tasks like understanding user intent or query classification can run on smaller, cheaper, or even open-source models. The heavier compute is reserved for complex, longer-running tasks. The variety of releases, he said, reflects that range of customer needs.
On where Qwen stands against ChatGPT or Google Gemini, Lin was straightforward: “Our biggest strength is a really similar level of intelligence at a fraction of the cost.” He used the local petrol analogy: most people do not need RON 97, and RON 95 is good enough. Choong added that the company’s position is cost-effectiveness, not being the cheapest or the most expensive, but enabling customers to get the most from the lowest possible spend.
Data sovereignty did not feature in Alibaba Cloud’s prepared remarks. When it was raised as a question at the end of the session, Lin acknowledged the omission directly: “That’s a really good question, pointing out our blind spot that is very important, which we forgot to mention.”
He confirmed that the company has purpose-built infrastructure in Malaysia to address data residency requirements, with local inference capabilities serving financial services customers and government agencies that need data to remain within the country. The infrastructure is designed so that AI inference is served locally, keeping customer data in Malaysia rather than being routed externally.
Choong shared a number that did not appear in the press release: Alibaba Cloud has onboarded more than 300 local system integrators, resellers, distributors, and partners in Malaysia since entering the market. The network spans enterprise and SME customers. Over the past one to two years, Alibaba Cloud has been running AI-focused training programmes for these partners, on the logic that trained partners go on to train their own customers.
“Malaysia remains a strategic key market for us in Southeast Asia,” Choong said. “This expansion is a direct response to surging customer demand as local businesses scale cloud-native operations and integrate AI at scale.”
The other named local partnership in the press release is TNG Digital, which is using Alibaba Cloud’s unified data platform to enhance search and recommendation capabilities for TNG eWallet.
Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology events, click here for more information.
Tech Wire Asia is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。