Malaysia's Overload: AI Data Centers Threaten the Power Grid and Water Resources
The rapid development of artificial intelligence infrastructure in Malaysia is creating critical pressure on the country's key resources. My analysis shows that the nation, aiming to become a regional hub for data centers (DCs), faces serious challenges in energy supply, water usage, and staffing.
Numbers That Cannot Be Ignored
According to my data, a single data center with a capacity of 100 MW consumes approximately 4.2 million liters of water per day — a volume comparable to the water supply of a small town. Meanwhile, about 80% of all operational capacity in the country is concentrated in the state of Johor, creating a local imbalance in infrastructure load.
A particularly alarming forecast indicates that by 2029, the total capacity of data centers in Malaysia could grow to 3–4 GW. For comparison, this is equivalent to the capacity of several large nuclear power plants. Such growth rates call into question the national power grid's ability to handle peak loads.
Water and Staffing Crisis
Beyond electricity, water supply is becoming a key issue. Data center cooling systems require enormous volumes of water, especially in Malaysia's tropical climate, where the efficiency of air cooling is reduced. During periodic droughts, this could spark conflict between industrial and domestic consumers.
The staffing issue is also intensifying. Building and maintaining such facilities requires qualified engineers, technicians, and AI infrastructure specialists, of which the country currently has an insufficient supply. This leads to an overheated labor market and rising wages in the sector, which could negatively impact other areas of the economy.
My professional opinion: Malaysia risks repeating the mistakes of other Asian economies that, in the pursuit of AI investments, underestimated infrastructure costs. Without coordinated government policy to modernize power grids and implement water-saving technologies, the country could face a systemic crisis by the end of the decade. Investors in Malaysian data centers should more carefully assess the risks associated with resource provision.