Malaysia AI Data Centre Policy: 2026 Analysis Guide

Malaysia’s data centre approvals are becoming more selective. BMI, a unit of Fitch Solutions, linked this shift to a policy announced by Prime Minister Datuk Seri Anwar Ibrahim in February 2026. Under this approach, proposals for non-AI-related data centres in Malaysia have been halted since 2024, reinforcing a preference for larger AI facilities that require more capital. The intent, according to BMI, is to prioritise approvals for bigger projects while advancing Malaysia’s AI capabilities. In practice, this also raises barriers to entry, especially in Johor, and increasingly favours large and established players over new entrants.

This selectivity is already shaping what gets built. BMI described Malaysia’s capacity pipeline as virtually entirely AI-focused, with estimates pointing towards 4.6GW of capacity that is planned or under construction. The same analysis framed Southeast Asia as a bifurcated market, where other countries’ pipelines are more tied to enterprise needs and broader digital transformation rather than AI-heavy workloads. BMI also expects displaced non-AI demand to be pursued by Indonesia, Thailand, and the Philippines, especially as investors look for alternatives amid saturation and infrastructure bottlenecks in primary markets.

Why the Policy Is Also About Infrastructure and Sustainability

The Malaysia AI Data Centre Policy debate is not only about which projects are approved. It is also about what the grid and utilities can realistically support. TechWire Asia cited analysis from Area Market Intelligence that points to grid connection delays, rather than electricity generation capacity, as the bottleneck for new developments. Separately, BMI said Malaysia was rejecting nearly 30% of data centre proposals that failed to demonstrate responsible water and power consumption practices. That rejection rate signals that growth is being tied more explicitly to environmental standards, not just headline investment announcements.

Industry voices are also cautioning against treating rapid expansion as the only goal. The New Straits Times reported experts urging Malaysia to be more selective, given pressure on electricity, water, and land resources. Muhammad Hazim Mohd Halimi of Thinker Engineering Sdn Bhd said electricity can account for around 60% to 70% of operating expenditure, with the rest going to connectivity, cybersecurity, staffing, and overheads. He argued it makes sense to prioritise projects that bring stronger technological spillovers and more meaningful contributions to the domestic ecosystem. He also warned that a heavy tilt toward AI-specific capacity, without sufficient underlying demand, could create an imbalance.

For local companies, the shift may split the market into two tiers. AIMS Data Centre said Malaysia has quietly been rejecting new applications since mid-2024 unless they show a clear AI or high-tech purpose, and noted a divide between hyperscale AI facilities for global players and managed enterprise infrastructure serving Malaysian businesses. AIMS also stated the sector is forecast to support USD 34 billion in economic output by 2030, while naming global players such as Google, Microsoft, Amazon, and DayOne as having committed significant capital to large-scale facilities. Asia Society added a different risk lens: it said much of Malaysia’s AI engagement comes via foreign hyperscaler-funded data centres, and that data centres tend to generate limited spillover compared with other industrial facilities. It cited APDCA estimates that the Malaysian AI industry is projected to create 30,900 jobs annually by 2030, while also noting research that data centres employ thousands during construction but barely 200 during operation.

What changed under Malaysia’s newer data centre approval approach?

BMI said Malaysia formalised a policy in February 2026 that prioritises larger AI data centres. Proposals for non-AI-related data centres have been halted since 2024.

How much data centre capacity is in Malaysia’s pipeline, according to BMI?

BMI estimates point to 4.6GW of capacity in Malaysia that is planned or under construction. BMI also described the pipeline as virtually entirely AI-focused.

How strict is Malaysia on power and water responsibility in data centre proposals?

BMI said Malaysia was rejecting nearly 30% of data centre proposals that failed to demonstrate responsible water and power consumption practices. This frames growth as conditional on environmental standards.

What is the main infrastructure bottleneck mentioned in reporting on the Malaysia AI Data Centre Policy?

TechWire Asia cited Area Market Intelligence analysis stating that grid connection delays, not electricity generation capacity, are the bottleneck for new developments.

What job and operations reality does Asia Society highlight about data centres?

Asia Society cited APDCA estimates that the Malaysian AI industry is projected to create 30,900 jobs annually by 2030. It also cited research saying data centres employ thousands during construction but barely 200 during operation.