AI Helps Singapore Semiconductor Firms and Data Centres Slash Water Use
Source: CNA
Singapore's semiconductor manufacturers and data centre operators are deploying AI-powered water management to tackle their massive water footprint. UMC's Pasir Ris facility now recycles nearly 70% of its water, recovering 3.9 million cubic metres annually, while OVHcloud uses predictive AI to cut cooling consumption by up to 30%.
Singapore's semiconductor manufacturers and data centre operators are deploying AI-powered water management systems to dramatically reduce their environmental footprint. UMC's wafer fabrication plant in Pasir Ris now recycles nearly 70% of its water, recovering 3.9 million cubic metres annually — enough to fill 1,560 Olympic-sized swimming pools — while French cloud provider OVHcloud is using predictive AI to cut cooling water consumption by up to 30%.
Wafer fabrication is among the most water-intensive manufacturing processes on the planet, requiring ultra-pure water across roughly 1,000 individual manufacturing steps per chip. Singapore's status as a global semiconductor hub means the industry's water consumption is a significant national concern. Data centres, whose expansion is being propelled by surging AI demand, are equally thirsty — they use vast quantities of water for server cooling. The national water agency PUB has allocated S$12 million specifically for water efficiency research targeting these two sectors.
UMC achieves its 70% recovery rate through a sophisticated wastewater segregation system using 20 different pipeline channels, each carrying water of different contamination levels that gets treated separately and reused for specific operations — from precise process steps to cooling towers. The company is now exploring membrane-based treatment technologies including reverse osmosis and electrodeionisation to push recovery rates even higher. Meanwhile, OVHcloud has deployed smart sensors across its data centres that feed real-time server, rack, and cooling data into a central analytics platform. AI models predict cooling demand based on load patterns, adjusting water injection throughout the day rather than running cooling systems at full blast continuously.
The combination of AI-driven optimisation and advanced water treatment gives Singapore's industrial sector a replicable playbook for managing water scarcity. As AI adoption drives further expansion of both semiconductor fabrication capacity and data centre infrastructure, these efficiency gains become critical to keeping the industry's growth sustainable. OVHcloud estimates its approach could reduce total water consumption by 30% and power usage by up to 50%, while exploring integration with local weather data for even finer-grained control.
Why it matters for Singapore: The city-state's AI ambitions rest on a physical foundation of chips and data centres that consume enormous water resources. These AI-powered water management techniques demonstrate that Singapore can grow its digital economy without outstripping its water supply — a tension that will only intensify as AI demand accelerates.