Free subscribers! You’re watching a short preview of the full training episode. Upgrade to paid below to instantly unlock this video, all weekly videos and the entire collection of past 350 trainings. You can unsubscribe at the bottom of the page.
The Tech Learning Academy is now 22,000 members strong. We are 100% member supported with no ads or corporate agenda.
When you sit at your desk, type a prompt into a window, and ask a chatbot to write a poem or generate an image of a sunset, the magic happens instantly. But that magic comes with a massive, physical cost.
Hundreds of miles away, millions of transistors are flipping to process your request. When transistors flip, they generate friction and massive, compounding amounts of heat. If that heat isn’t removed, the silicon literally melts. To cool down the most advanced technology in human history, the industry relies on the oldest and most fundamental mechanism on Earth: water.
Artificial intelligence is incredibly thirsty, and we are finally addressing the elephant in the room. Here is the reality of the AI water problem, and how Google is attempting to solve it.
The Scale of the Thirst
A recent study revealed a staggering parallel: AI technology consumes roughly the same amount of water every year as the entire global population drinks from plastic bottles. Every single prompt is a drop, and those drops are quickly forming oceans.
The scale of this operation is only expanding. Alphabet (Google) recently announced a goal to raise $80 billion through stock sales specifically to fund their AI build-out. We are pouring oceans of capital and actual oceans of water into our silicon.
Naturally, the public is pushing back. A Gallup poll showed that more than 70% of Americans actively oppose the construction of a data center in their local area. Half of those surveyed cited the severe impact on environmental resources as their primary reason for opposition, with 18% pointing directly at excess water usage as their main concern. Furthermore, some researchers point out that prior estimates of AI water usage were misleadingly low because they failed to include indirect water use throughout the entire supply chain.
The Thermodynamics Trap: Why Not Use Air?
The obvious question is: why use water at all? Why not just use giant fans to cool the servers with air?
According to Vikash Kohli, Vice President of Global Infrastructure at Google, thermodynamics is a cruel master. Air is a terrible conductor of heat, but water is brilliant at it. Utilizing water to cool data centers actually reduces overall energy consumption by about 10% compared to using air alone. It is a brutal but necessary trade-off—we save electricity, but we spend finite water.
To gain some perspective, Kohli noted a fascinating statistic: all of the data centers in the United States consume less than 1% of the water that Americans spray onto their residential lawns every single year. We are outraged by the computers, yet perfectly happy watering our grass.
Google’s 2030 Blueprint
So, how do we build the future without draining the present? Google has published a pledge featuring five specific commitments regarding their water usage:
By 2030, Google promises to replenish more water than their data centers are actually consuming, making them self-sustainable.
They are actively abandoning tap water and identifying alternative sources, such as using reclaimed fresh water to cool massive facilities in Georgia.
The company announced a $17 million investment to fund new local water stewardship projects across different states.
Google commits to continuing to report its annual water use to maintain public transparency.
According to Ben Towsant, Google’s Global Head of Infrastructure and Sustainability, the goal is to establish a public blueprint. If another tech giant wants to build a massive facility in your hometown, citizens can point to this blueprint, look them in the eye, and demand they prioritize the local watershed.
Water is the absolute prerequisite for all biological life, and it is now the prerequisite for synthetic intelligence. As we build brilliant new minds of glass and silicon, we must teach the machines how to respect the water cycle.
