OpenAI has established two computing infrastructure centers in Malaysia: Why has Southeast Asia shifted from a 'backup node' to a primary battlefield for AI?

Australian AI infrastructure company Firmus announced on September 8 that it has signed a multi-year compute services agreement with OpenAI, under which it will supply computing power to OpenAI from two data centers located in Malaysia. The parties did not disclose the contract value or the megawatt capacity each facility will provide. Thus, while this agreement confirms OpenAI as a major customer, it does not permit further inference about any undisclosed order volume. Following the signing, Firmus stated that its total contracted capacity across all customers exceeds 900 megawatts. The company currently operates two AI-focused data centers—in Australia and Singapore—and is advancing five additional projects across the Asia-Pacific region. Firmus plans large-scale deployment of NVIDIA’s Vera Rubin systems across the region. 'Contracted capacity' does not equate to operational servers. Data centers require completion of construction, electrical grid connection, cooling infrastructure, network setup, and equipment installation; customers typically also conduct acceptance testing. Some portion of the 900 MW may still be in planning or construction phases. Malaysia has attracted numerous data center projects in recent years, largely because land and construction costs are comparatively lower than in Singapore—while still enabling connectivity to Southeast Asian subsea cables and regional enterprise markets. After Singapore imposed stricter energy-efficiency and resource-use requirements for large-scale data centers, some new projects began shifting toward Malaysia. Yet relocating facilities does not eliminate resource constraints. 900 MW approaches the output scale of a large power plant. Servers demand not only continuous electricity but also substantial cooling water—or expensive closed-loop cooling systems.

Within just a few years, billions of dollars have poured in—why has Malaysia become a high-ground for data center investment in Asia?

The report notes that Johor Bahru’s total data center supply has reached 1.6 GW—including facilities under construction, committed, or in early planning stages. Typically, data center capacity is measured by its electricity consumption. If all planned data centers across Asia come online, only Japan and India would surpass Malaysia in total capacity. Until now, Japan and Singapore lead Asia in operational data center capacity. This index does not provide disaggregated data on operational data center capacity in China and therefore excludes the Chinese market. Rising AI Compute Demand Traditionally, the vast majority of investments in data center infrastructure and storage flowed into mature markets such as Japan, Singapore, and Hong Kong. However, according to a research report released by global data center provider EdgeConneX, the COVID-19 pandemic accelerated global digital transformation and cloud adoption—sparking surging demand for cloud services in emerging markets like Malaysia and India. Murphy stated: “As user demand surges for video streaming, data storage, and various internet- or mobile-delivered services, demand for data centers grows correspondingly.” Meanwhile, robust demand for AI services requires specialized data centers capable of handling the massive data volumes and computational power needed to train and deploy AI models. They note that although many AI-system data centers will be built in mature markets like Japan, emerging markets—with their distinct advantages—will also attract substantial investment. AI data centers require abundant space, energy, and cooling water. Therefore, compared to smaller cities or nations with constrained resources—such as Hong Kong and Singapore—emerging markets like Malaysia, where energy and land are relatively inexpensive, hold greater competitive advantage. The Singapore Spillover Effect Malaysia’s pro-data-center policies have made it an exceptionally attractive market. In 2023, the Malaysian government launched a 'Green Lane' initiative aimed at streamlining electricity approval processes—reducing data center delivery timelines to 12 months. Yet another key catalyst in recent years has been Singapore’s restrictive policies. Although Singapore’s talent pool, business reputation, and fiber-optic connectivity make it an appealing location for data center investment, its large-scale energy and water consumption led Singapore to slow data center capacity expansion starting in 2019. Consequently, over the past several years, a significant volume of data center investment and construction plans has shifted from Singapore to neighboring Johor Bahru. More recently, Singapore’s stance has shifted: it has formulated a roadmap to increase its data center capacity by 300 MW—on the condition that more projects meet green efficiency and renewable energy standards.

Iran War Boosts Malaysian Data Center Appeal, SD Guthrie Says

1 day ago

SD Guthrie links rising data center investment to regional stability while forecasting palm oil price increases due to Indonesian harvest constraints. The company aims to utilize its vast landbank for industrial projects alongside its core agricultural business.

bloomberg

2026-10-26

Iran War Boosts Malaysian Data Center Appeal, SD Guthrie Says

Iran War Boosts Malaysian Data Center Appeal, SD Guthrie Says

FAQ