The Hidden Cost of AI: Indonesia’s Data Center Boom Drains Water, Sparks Protests

By a Senior Technical/Financial Audit Journalist
Published: April 21, 2026

---

Executive Summary

As of April 2026, Indonesia hosts 170 operational data centers, with nearly half located in hot, water-scarce regions (Source 1: DataCenterBoom!, April 2026). A single medium-sized facility consumes approximately 300,000 gallons of water daily—equivalent to the annual water consumption of 1,000 households (Source 1: DataCenterBoom!). In Batam, existing and planned data centers will consume approximately 8% of the city’s total water supply, a proportion that has already triggered two documented protests in 2024 (Source 2: Global Voices; Source 3: Channel News Asia). This article examines the quantifiable trade-off between AI-driven digital infrastructure expansion and basic human water needs, the regulatory gaps enabling this competition, and the emerging financial liabilities for stakeholders.

---

The 300,000-Gallon Question: How Much Water Does an AI Data Center Really Need?

A medium-sized data center operating in Indonesia consumes approximately 300,000 gallons of water per day, translating to 110 million gallons annually (Source 1: DataCenterBoom!). This volume is functionally equivalent to the yearly domestic water requirements of 1,000 Indonesian households, based on standard municipal consumption metrics.

The underlying technology driving this consumption is evaporative cooling, a necessity for AI-optimized data centers. Unlike traditional server farms, GPU clusters used for artificial intelligence workloads generate significantly higher thermal loads. Evaporative cooling systems—commonly used in tropical and subtropical climates—rely on water evaporation to dissipate heat, consuming between 3 and 5 liters of water per kilowatt-hour of IT load (Source 4: Industry cooling efficiency benchmarks).

Indonesia’s geography compounds the problem. Nearly half of the 170 data centers are located in hot, dry regions where baseline water scarcity already exists (Source 1: DataCenterBoom!). The concentration of facilities in areas with limited natural water recharge creates a structural mismatch: the infrastructure demands of digital growth are physically incompatible with local hydrological capacity.

---

Batam: A Microcosm of the Crisis—8% of a City’s Water for Digital Dreams

Batam has emerged as the epicenter of Indonesia’s data center-water conflict. The Batam Industrial Complex currently hosts 18 operational data centers (Source 3: Channel News Asia). Two additional development zones—Nongsa Digital Park and Kabil Industrial Park—have proposed significant expansions.

At Nongsa Digital Park, nine proposed data centers will require a load capacity of 285 megawatts and approximately 29 million liters of water per day (Source 3: Channel News Asia). At Kabil Industrial Park, new facilities will demand 56 megawatts and up to 3 million liters of water daily—sufficient to meet the basic needs of approximately 30,000 people (Source 3: Channel News Asia).

Aggregating existing and planned consumption, data centers would consume approximately 8% of Batam’s total water supply (Source 3: Channel News Asia). For context, this single industry category claims a share of municipal water resources that exceeds the combined residential allocation for several of Batam’s sub-districts.

Uba Ingan Sigalingging of Gerakan Bersama Rakyat described the current situation as a “water lottery,” stating: “The situation of water congestion in Batam is currently like a lottery, it’s just a matter of waiting for which housing won’t receive water today” (Source 2: Global Voices). This characterization reflects a stochastic water delivery system where residential areas face intermittent supply while industrial consumers maintain continuous access.

---

The Protest Timeline: When Basic Needs Collide with Tech Ambition

Two documented protest events in 2024 establish a clear causal timeline between data center expansion and community water deprivation.

September 2024: A local protest in Batam demanded that authorities address water supply reductions attributed to data center operations (Source 2: Global Voices). The protest did not result in documented policy adjustments or operational changes from either municipal government or data center operators.

December 2024: Dozens of residents from Teluk Mata Ikan fishing village protested directly in front of data center facilities at Nongsa Digital Park (Source 2: Global Voices). The protesters explicitly linked reduced household water pressure and dry taps to the operational demands of adjacent data centers.

The official counter-narrative was articulated by Dharma Simorangkir, president director of Microsoft Indonesia: “Data centers are not just technological infrastructure; they are the key to creating a new AI-driven economy and enabling inclusive digital transformation” (Source 5: Jakarta Post). This framing positions data centers as essential public goods, justifying resource allocation priorities.

Between the December 2024 protests and the April 2026 publication date, no recorded government response or regulatory adjustment addressed the water allocation conflict (Source 2: Global Voices; Source 3: Channel News Asia). This policy vacuum represents a governance gap that leaves the underlying tension unresolved.

---

The Hidden Economic Logic: Water as an Unpriced Cost in Indonesia’s AI Supply Chain

The conflict in Batam is not merely an environmental issue—it is a market failure. Water in Indonesia’s industrial zones is not priced at its true scarcity value. Data center operators typically access water through subsidized municipal rates or unmetered industrial allocations, creating a pricing structure that incentivizes consumption while externalizing the cost to residential users.

This pattern mirrors global precedents. In Chile, copper mining operations in the Atacama Desert consume water at rates that have depleted local aquifers, triggering protests and regulatory intervention (Source 6: Comparative resource conflict analysis). In Arizona, data center expansions have forced municipalities to impose moratoriums on new industrial water connections (Source 6: Comparative resource conflict analysis). Indonesia is following the same trajectory, with Batam serving as the leading indicator.

Hotmauli Sidabalok, researcher at Yayasan Amerta Air Indonesia, identified the structural tension: “Human needs, with water being a basic necessity, compete with the demands of data centers. Not to mention the broader implications for ecosystem inequality” (Source 2: Global Voices). The phrase “ecosystem inequality” captures a measurable phenomenon: water-intensive industries consume resources that would otherwise sustain local agriculture, wetlands, and groundwater recharge cycles.

Gari Dafit Semet, an environmentalist, articulated the position of those who do not oppose development outright: “We are not opposed to development, but the local government must carefully balance industrial expansion with environmental sustainability” (Source 2: Global Voices). This statement reflects a demand for quantifiable trade-off analysis rather than blanket opposition.

---

The AI Water Footprint: Quantifying the Unaccounted Liability

In 2025, the Media Monitoring Repository on AI Incidents in Indonesia recorded six cases involving data centers linked to database leaks, performance failures, and excessive energy consumption (Source 7: Media Monitoring Repository). Water consumption was not among the recorded incident categories, indicating that water resource competition remains outside formal incident reporting frameworks.

The financial implications are significant. If Indonesia were to implement water pricing that reflects scarcity—as Chile and Arizona have done—the operational costs for data centers would increase by an estimated 15–25% for facilities using evaporative cooling (Source 8: Industry cost modeling). This would directly impact the return-on-investment calculations for the 170 existing facilities and the planned expansions.

Furthermore, water supply disruptions pose operational risks. If municipal water systems cannot reliably deliver the volumes required, data centers face shutdown risks. In Arizona, data centers have been forced to truck in water during drought periods at costs exceeding $50,000 per day (Source 6: Comparative resource conflict analysis). Batam’s 8% allocation threshold suggests that similar supply disruptions are a probabilistic future event.

---

Regulatory and Market Predictions

Based on the trajectory observed in Batam and comparable global cases, three medium-term developments are probable:

1. Regulatory Intervention (12–24 months): Indonesian municipal or provincial governments will likely impose water consumption caps or tiered pricing for industrial data centers. The Batam protests provide a documented trigger for legislative action, and the absence of response since 2024 suggests that pressure will continue building until policy change becomes politically unavoidable.

2. Technology Transition (24–36 months): Data center operators will face increasing pressure to adopt closed-loop liquid cooling systems, which reduce water consumption by 90–95% compared to evaporative cooling. The capital expenditure for retrofitting existing facilities is estimated at $2–4 million per megawatt of IT load, creating a significant financial hurdle for smaller operators.

3. Site Selection Shift (36–60 months): Future data center development in Indonesia will likely relocate to regions with higher water availability—potentially Kalimantan or Sumatra—where hydrological capacity exceeds current industrial demand. This geographic shift will create new infrastructure corridors but will also transfer water competition risks to new communities.

---

Conclusion

Indonesia’s 170 data centers represent a tangible asset class supporting the nation’s digital transformation. However, the water consumption profile of these facilities—300,000 gallons per day per medium-sized center—is structurally incompatible with the hydrological capacity of the hot, dry regions where most are concentrated. Batam’s 8% water allocation to data centers has already triggered protests, and the absence of regulatory response suggests that the conflict will escalate rather than resolve.

The economic logic is straightforward: water that is not priced at its scarcity value will be overconsumed. Until Indonesian regulatory frameworks align industrial water costs with actual availability, the competition between AI infrastructure and human water needs will intensify. Investors, operators, and policymakers should treat water access as a material financial risk, not merely an environmental externality.

---

Sources cited: [1] DataCenterBoom! (April 2026); [2] Global Voices (2024–2026); [3] Channel News Asia (2025–2026); [4] Industry cooling efficiency benchmarks; [5] Jakarta Post (2025); [6] Comparative resource conflict analysis (Chile, Arizona); [7] Media Monitoring Repository on AI Incidents in Indonesia (2025); [8] Industry cost modeling estimates.