Malaysia’s AI boom has a hidden cost: Water — Mohd Firdaus Abdullah

SEPTEMBER 11 — Malaysia wants to become an AI nation. But every new cloud campus, hyperscale facility and AI factory rests on something far less futuristic: water.

The country’s data-centre boom is accelerating. In the first half of 2026, data-centre and cloud-computing projects accounted for RM95.8 billion in approved investments, close to 44 per cent of Malaysia’s total approved investment. MIDA says around 4.6 gigawatts of capacity is planned or under construction. Johor, with its proximity to Singapore and large industrial land banks, sits at the centre of this expansion.

This is good news for Malaysia’s ambition to move up the digital value chain. But it also raises a question that investment announcements rarely answer: how much water will this new digital economy require, where will that water come from, and who gets priority when supply becomes tight?

The danger is not that data centres have already caused a water crisis. That claim would be too crude. The more important problem is that the infrastructure decisions being made today are locking in tomorrow’s demand.

Johor’s own official figures show why this matters. In September 2025, the state government told the Johor assembly that its 47 water-treatment plants had a combined design capacity of 2,352 million litres a day. The statewide reserve margin stood at 14.5 per cent. In Greater Johor Bahru and Kulai, where much of the data-centre expansion is concentrated, design capacity was 1,274.1 million litres a day, while average production had already reached 1,129.1 million litres a day. That left a reserve margin of just 11.4 per cent.

Malaysia wants to become an AI nation. But every new cloud campus, hyperscale facility and AI factory rests on something far less futuristic: water. — Picture by Sayuti Zainudin

The state’s own analysis projected water demand in Johor Bahru and Kulai to reach 1,600 million litres a day by 2030 and 1,800 million litres a day by 2040, driven by housing, industry, commercial growth and data centres.

These figures do not mean Johor is destined to run out of water.

They mean its room for error is narrowing.

The state is already spending and planning to expand supply. Johor has proposed or pursued new river-bank storage schemes, barrages, treatment facilities and water-transfer infrastructure. Years before the current AI boom reached its present scale, the state assembly was already discussing the water requirements of Sedenak Tech Park: its first two phases alone were expected to require 19.62 million and 31.44 million litres a day respectively.

That history matters. Data-centre water demand is not merely a question of how efficiently individual servers are cooled. It is a question of cumulative demand across an entire basin and distribution system.

Malaysia has recognised part of the problem. The federal Guideline for Sustainable Development of Data Centres recommends that new facilities avoid water-stressed areas with a Water Stress Index of 0.8 or above in Peninsular Malaysia. It recommends a design Water Usage Effectiveness, or WUE, of 2.2 cubic metres per megawatt-hour or lower, annual disclosure of actual WUE, and gradual improvement towards 2.0 or lower. It also allows operators to use reclaimed or reused water.

Johor has begun moving in that direction. In August 2025, Indah Water Konsortium and Johor Special Water announced that up to 12 million litres a day of reclaimed water could be supplied to two data centres in Johor, reducing their dependence on treated drinking water. This is precisely the kind of infrastructure innovation Malaysia should expand.

But efficiency is not the same as accountability.

A data centre can record an impressive WUE and still impose significant pressure on a water system if the facility is enormous, if several facilities cluster in the same catchment, or if demand rises during a drought. A national average also tells residents little about what is happening in their own basin.

Malaysia therefore needs a more demanding principle for the next phase of its AI expansion: no large digital infrastructure project should be considered sustainable merely because it satisfies an efficiency ratio.

First, water disclosure should become much more transparent. Large data centres should publish not only WUE but total annual withdrawal, monthly consumption, peak daily demand, water source, the proportion of potable and reclaimed water used, and the catchment from which supply is drawn. These figures should be accessible through a national data-centre water registry.

Second, reclaimed water should become the default for cooling wherever technically feasible, particularly for hyperscale facilities in water-sensitive areas. Malaysia’s current guideline says operators “may” use reclaimed or reused water. That language was a reasonable starting point. It should not be the end point.

If wastewater can be treated to a standard suitable for industrial cooling, potable water should not automatically remain the cheapest or easiest option. Drinking-quality water is a public resource whose highest social value is not necessarily found inside a cooling tower.

Third, Malaysia needs clear drought rules for data centres before the next serious drought arrives. What happens when river flows fall sharply, reservoirs decline and households are asked to conserve water? Which users face restrictions first? What minimum contingency storage or alternative supply should a hyperscale facility maintain?

These questions should be answered in policy, not improvised during an emergency.

Domestic and essential public needs should have explicit priority. Large industrial users should know in advance the thresholds at which allocation will be reduced, and they should be required to demonstrate contingency arrangements.

Fourth, Malaysia should judge data-centre projects by the value they create relative to the resources they consume.

The debate has already shifted in this direction. MIDA now stresses that the country wants high-value, AI-focused digital investments, with stronger local supply chains, high-skilled employment and collaboration with universities and technical institutions.

That principle should be extended to water.

If a data-centre cluster requires the state to accelerate new reservoirs, treatment plants, pipelines or alternative-water facilities, Malaysians should be able to see the return: skilled jobs, local procurement, research capacity, AI infrastructure, tax contribution and meaningful technology transfer.

The relevant measure is no longer simply how many billions of ringgit are announced or how many megawatts are built. It is how much long-term national value is created for every unit of land, electricity and water committed.

This matters beyond Malaysia.

The AI industry is often described in language that makes it appear almost immaterial: clouds, models, tokens and virtual infrastructure.

But the cloud is not weightless.

Artificial intelligence ultimately sits on concrete, steel, transmission lines, substations, cooling equipment, treatment plants, rivers and pipes.

For countries across the Global South seeking a share of the AI economy, this physical reality will become increasingly important. They will be asked to compete for investment while simultaneously managing urban growth, industrialisation, climate variability and basic public services.

Malaysia can offer a different model.

It does not need to choose between AI and water. It needs to refuse a development model in which the costs of digital growth remain invisible until scarcity makes them political.

The country has already taken important steps: sustainability guidelines, screening of data-centre projects, WUE requirements and reclaimed-water projects. But the next step must be stronger: basin-level planning, transparent disclosure, enforceable drought priorities and a presumption in favour of alternative water for large cooling loads.

Malaysia’s AI boom should continue.

But its success should not be judged by how rapidly the country can fill industrial parks with servers.

It should be judged by whether Malaysia can build a digital economy without making households, ecosystems and future generations absorb its hidden water bill.

* Mohd Firdaus Abdullah (PhD) is a senior lecturer at the Centre for Research in History, Politics & International Affairs, Universiti Kebangsaan Malaysia. His expertise is on history of Malaysia particularly water development, water history and water politics in Malaysia.

** This is the personal opinion of the writer or publication and does not necessarily represent the views ofMalay Mail.