This Asian data center hub is grappling with the massive costs of AI: energy and water

Published Mon, Aug 18 202512:34 AM EDTUpdated Mon, Aug 18 20255:33 AM EDT

Key Points

Malaysia's Johor has a total planned data center capacity of 5,800 megawatts, according to DC Byte.
Data center projects have created economic opportunities and jobs for the city.
But there are signs the industry is pushing the limits of the state's capacity and natural resources.

The artificial intelligence boom has brought with it hundreds of billions of dollars in investments and promises of economic growth . But the infrastructure required is demanding massive amounts of energy and resources. One lesser-known example of that dilemma can be found in the southern tip of Malaysia, which has quietly become one of Southeast Asia's fastest-growing data center hubs amid the heightened compute demands of AI. The country's state of Johor — with a population of about 4 million people — has attracted billions' worth of projects for such data centers in recent years , including from many of the world's largest technology firms, such as Google , Microsoft and China's ByteDance . Backers of those projects have been drawn by Johor's cheap land and resources, proximity to the financial hub of Singapore, and government incentives. But though that has created new economic opportunities and jobs, there are signs the industry is pushing the limits of the state's energy capacity and natural resources, with officials slowing approvals for new projects. Energy needs and hurdles While Johor currently has about 580 megawatts (MW) of data center capacity, its total planned capacity — including early-stage projects — is nearly 10 times that amount, according to figures provided by data center market intelligence firm DC Byte. That energy capacity would be enough to power up to 5.7 million households an hour , according to calculations based on data from PKnergy . Meanwhile, though Johor accounts for the majority of Malaysia's planned data centers, other hubs in the country have been sprouting up. Kenanga Investment Bank Berhad, a Malaysian independent investment bank, has projected that the electricity use of the country's data centers will equate to 20% of its total energy-generating capacity by 2035. In the face of those growing demands, a Malaysian industry official told reporters in June that the country expects to add 6 to 8 gigawatts of gas-fired power, with total power consumption on track to increase 30% by 2030. Though the natural gas used in these power stations burns cleaner than coal — which accounted for more than 43% of Malaysia's electricity in 2023 — reliance on it for future data center expansion could clash with the country's plan to achieve net-zero emissions as early as 2050. Another critical challenge is water, which is used by data centers in large quantities to cool down electrical components and prevent overheating. It's been estimated that an average 100 MW data center uses about 4.2 million liters of water per day — the equivalent of supplying thousands of residents . It's therefore no surprise that Johor, which has experienced several supply disruptions and already relies on neighboring Singapore for a sizeable amount of its treated water, is reportedly in the process of building three new reservoirs and water treatment plants. Global picture Data centers are the backbone of the digital world, hosting the information and computing resources that power everything from e-commerce to social media to digital banking, and increasingly, generative AI models. Demand and investor appetite for such centers have never been higher, given the massive computing power needs of AI , with Johor serving as just one example of the industry's growth and the energy and water challenges that come with it. According to a May report by the International Monetary Fund, electricity used by the world's data centers had already reached the levels of Germany and France in 2023, soon after the launch of OpenAI's groundbreaking ChatGPT AI model. Meanwhile, some researchers have estimated that AI-related infrastructure could consume four to six times more water than Denmark by 2027. The industry's growth is expected to continue to accelerate, though projections of future capacity vary widely. One thing that is clear is that data center construction is struggling to keep up with demand in light of power constraints and permitting delays, according to DC Byte. In response, some governments have been working to speed up approval processes and bring new and cheap energy online, with some environmentalists warning such moves could clash with global net-zero goals. The United States — the world's largest data center market— has exemplified that dynamic. U.S. President Donald Trump recently launched " America's AI Action Plan ," calling for streamlined permitting and the removal of environmental regulations to speed up the development of AI infrastructure and the energy needed to power it. A June analysis from Carnegie Mellon University and North Carolina State University projected that by 2030, Americans' electricity bills are on track to rise 8% and greenhouse gas emissions from power generation 30% as a result of growth in data centers and cryptocurrency mining. Resource solutions? Malaysia, for its part, has signaled its desire to rein in the data center industry's energy and resource use. The government plans to launch a "Sustainable Data Centre Framework" by October, Tengku Zafrul, investment, trade and industry minister, said in a post on X in July. To meet growing power needs, officials have also been approving more renewable energy projects , while also exploring the potential use of nuclear energy . As for water, higher water tariffs were placed on Johor's data centers earlier this month, with the government pushing for the industry to shift to using recycled wastewater . Notably, some newer data centers don't rely on any water for cooling. Regionally, concerns about resource-intensive data centers are nothing new. In 2019, Singapore cracked down on the industry, imposing a three-year moratorium on new data centers in order to stem power and water usage. It was after that crackdown that the industry began its major shift to the friendlier regulatory environment of Johor. Singapore ended its moratorium in 2022 and launched its "Green Data Centre Roadmap ," aimed at optimizing energy efficiency and adopting green energy for data centers. However, according to data from DC Byte, growth in the city-state remains tempered , especially when compared with Malaysia. In this way, Singapore shows how crackdowns on the data center industry can lead to spillover to less-regulated markets. As few international guardrails are in place, environmentalists and organizations like the United Nations Environment Programme have been calling for global legislation. "There are no unavoidable AI uses, and whether we move towards net-zero emissions is a choice," Jonathan Koomey , a leading independent researcher on the energy and environmental effects of information technology, told CNBC in an email. "There is no reason, in my view, why data center companies shouldn't power AI expansion with zero emissions power. There is also no reason to abandon climate goals because AI companies say their expansion is urgent."