Ecolab bets $7 billion that AI data centers can stop guzzling water
The water tech giant's acquisition of CoolIT Systems targets the trillion-gallon problem lurking behind every AI query you run.
Every time an AI model generates an image, answers a question, or writes code, water flows through a data center somewhere to keep the hardware from melting. Individual AI data center facilities can consume millions of gallons per day, and Ecolab CEO Christophe Beck says he’s spent $7 billion in six months to fix that.
The centerpiece of that spending spree: a $4.75 billion cash acquisition of CoolIT Systems, a company that builds direct-to-chip liquid cooling technology. The deal, announced in March 2026 and finalized on July 2, 2026, is Ecolab’s bet that the future of AI infrastructure runs through water management, not just power grids.
The trillion-gallon problem
Beck has projected that AI data centers could collectively consume over one trillion gallons of freshwater annually by 2027.
CoolIT’s technology works by running liquid directly to individual chips rather than blasting cold air through warehouse-sized rooms. The approach uses dramatically less water and enables circular cooling systems that recycle what they do use.
Ecolab’s goal is to make data centers “water-neutral,” meaning they’d return as much clean water to the environment as they consume. In 2025, Ecolab acquired Ovivo’s electronics unit for $1.8 billion, which focused on ultra-pure water for semiconductor manufacturing. The CoolIT deal extends that capability from chip fabrication into the data centers where those chips actually run.
The investment math
Ecolab, ranked No. 280 on the Fortune 500, is not a company that makes small bets by accident. The CoolIT acquisition is expected to double its addressable market in high-tech water applications from $5 billion to $10 billion. The company is targeting $4 billion in annual revenue from high-tech applications by 2030.