Dark Mode Light Mode

Subscribe to The Edit for regular how-to's and special invites to exclusive workshops!!

By pressing the Subscribe button, you confirm that you have read and are agreeing to our Privacy and Terms of Use.

Thirst Trap: The Hidden Water Cost of AI

AI seems so clean with its smooth fonts and invisible processes. But behind each web search or text-to-image creation is a physical infrastructure that consumes a vast amount of water. In Canada, training a large AI model can use over a million litres! At She Zine Mag, we acknowledge this impact and are taking action. Join us in exploring the hidden water cost of AI and our commitment to reclaiming what we use. Let’s create meaningful online content without harming the planet!
A woman cups her hands underwater A woman cups her hands underwater
Image Credit: Philia Daniel Farò

The Environmental Toll of AI in Canada—And What She Zine Mag Is Doing About It

Pixels, Power, and the Planet

AI seems so clean, doesn’t it? All smooth fonts and invisible processes. It writes our emails, corrects our grammar, suggests a playlist, maybe drafts your post outline when your brain is fried. It’s increasingly becoming integrated into the defaults of our devices, apps, and software. We might be using AI without even realizing it. 

We like to think of it as harmless. Gentle. Digital.

But just because you can’t see the smoke doesn’t mean there’s no fire.

Behind every grocery list, web search, or text-to-image masterpiece is a very physical infrastructure. One made of metal racks, loud fans, acres of servers, and—surprisingly—a shit ton of water.

That’s right. We’re talking about a full-blown environmental impact that’s getting harder to ignore. Especially in Canada. And especially if you, like us, are trying to make rad, meaningful, online things without cooking the planet in the process.

And while most of the world is panicking over whether ChatGPT is replacing artists, She Zine Mag is asking something else entirely:

How much water are we burning just to stay ‘clever cute’ online?
And more importantly:
How do we give it back?


The Hidden Water Crisis Behind “Clean” Tech

We tend to think of digital spaces as weightless. But AI is very much physical. Every prompt relies on data centres—huge buildings full of servers running hot and fast. And just like your laptop, those servers need to be cooled down. But unlike your laptop fan, this cooling process often involves millions of litres of water.

In Canada, the numbers look like this:

  • 1–1.5 litres of water per kilowatt-hour (kWh) consumed
  • Training a large AI model like GPT-4 = 1,000+ megawatt-hours, or 1 million+ litres of water
  • Even casual use—transcripts, summaries, image gen—can rack up 20–100 kWh/day for one small team

Even “casual” AI use (think: Midjourney art, auto transcripts, fancy SEO prompts) can use 20–100 kWh per day—per person, depending on how you’re using it.

TL;DR: AI may be digital, but it drinks like a landlocked whale.

This isn’t a glitch in the system—it is the system. The faster we demand our tools to respond, the harder the machines work behind the scenes, and the more water they burn cooling down the heat of all that performance.


The Indie Impact: She Zine Mag’s Annual Digital Footprint

We’ve tried to be extremely honest and transparent She Zine Mag regarding our AI use. We’re a tiny team doing big things, and AI helps us scale without soul-crushing burnout.

Here’s where we typically use it:

  • Image generation + editing
  • SEO and keyword support
  • Brainstorming prompts + wireframing
  • Research assistance

Here’s how it adds up:

Usage Scenario Approx. Prompts/Year Water Use (L) Carbon (CO₂/Year) Offset Goal (2x)
Conservative Use ~1,300 (3.56/day) ~1,950 L ~10 kg ~3,900 L
Heavy Use ~5,200 (14.24/day) ~7,800 L ~2,080 kg ~15,600 L

1 AI prompt = 1–1.5 litres of water.
That’s the water footprint of a single idea, headline, or draft. Multiply that by your entire content calendar—and now you’re looking at litres, not just likes.

And while we’re not pretending our impact is massive, we’re also not pretending it’s zero. Our footprint exists. So does yours.


⚡ ChatGPT vs. Google: Not Just a Smarter Search—A Thirstier One

I read a headline the other day that stated “ChatGPT is the new Google.”  It’s faster. It’s cleaner. It talks back.

But here’s what almost no one is saying:

Platform Water Use per Query Equivalency
Google ~0.0003 L per search[^1] Very low impact
ChatGPT ~1–1.5 L per prompt[^2] 3,000–5,000x more

That means one ChatGPT prompt can use as much water as 3,000 to 5,000 Google searches.

Crazy, right?

So while AI tools feel smart and seamless, they’re resource-hungry by design. You’re not just typing into a void or having a casual convo with your new digital friend—you’re firing up industrial systems with real environmental costs.

And unlike Google—which relies heavily on cached results—ChatGPT and its cousins generate every reply on the spot, running fresh inference cycles with every single interaction. It’s intelligent. It’s helpful. And it’s… wet.


What We’re Doing to Give It Back (No Greenwashing, Promise)

We don’t believe in greenwashing. We’re not “offsetting” in theory—we’re taking action in litres. We’re building a water-reclamation strategy that’s public, local, and unapologetically feminist.

Here’s some options for greener servers, conservation efforts, and options that we’re looking at as possible options for She Zine’s water-reclamation strategy:

1. Rain Gardens + Urban Water Projects

The installation of native plant rain gardens in Toronto neighbourhoods that experience flooding and runoff. These gardens absorb excess stormwater, restore pollinator habitats, and reduce pressure on aging infrastructure.

Canadian orgs to support:

These aren’t token gestures—they’re deeply local, highly effective, and scalable in cities that are already feeling the squeeze of seasonal flooding and infrastructure strain.


2. Trees That Intercept Stormwater

Tree planting in water-sensitive zones, especially urban areas that experience flash floods and erosion. Each tree can intercept thousands of litres of water per year.

Where to plant or donate:

Trees offer a triple win: cleaner air, cooler neighbourhoods, and less water runoff choking city drains during heavy rains.


3. Wetland and Freshwater Restoration

We’re exploring contributions to wetland reclamation projects, which act like nature’s kidneys—filtering water, storing carbon, and protecting communities from climate chaos.

Canadian orgs leading the way:

These projects aren’t just about biodiversity—they’re critical infrastructure for climate mitigation.


4. Greener Servers + Smarter Tech Use

We’re transitioning our hosting and AI workflows to green-powered Canadian servers, ideally those using Deep Lake Water Cooling—a system that draws cold water from Lake Ontario to reduce electricity use by 75% without evaporating water.

Green Canadian hosts:

What else can you do:

Use AI Mindfully

  • Ask yourself: Do I actually need this prompt?
  • Can Google (or other search engine) answer my question?
  • Don’t rely on a model to do your thinking—it’s bad for your brain and the environment.

Offset with Intention

  • Organize a local fundraiser for a tree-planting project
  • Install a rain barrel. Plant a pollinator garden. Fucking zine about it!

Support Indigenous-Led Water Protection

  • Indigenous communities are on the front lines of water defence
  • Land Back = Water Back = Real climate action

5. Mutual Aid and Micro-Offsets

Big changes are great. But small, local actions matter too.

We’re also planning to:

  • Install rain barrels in community gardens
  • Host pollinator garden pop-ups
  • Redirect merch profits to Indigenous-led water defence
  • Offer readers “Sponsor a Tree” or “Fund a Rain Garden” options with purchases and via our website

Groups to support or learn from:

Integrating these initiatives into our business model will take time but we will be publishing our quarterly water-reclamation report publicly available on our website so you can see the steps we’re taking and offer your own input for how we could do better. 

With new technology it’s a learning process for all of us and should be a collaborative effort. If you know of other Canadian programs or organizations that we should be considering or that you’d like to share with our audience, drop a link in the comments below.

What AI Companies Need to Do

While small publishers like us are hustling to reclaim what we use, the tech giants behind AI tools are wasting water by the millions of litres—without meaningful transparency.

Here’s what they should be doing instead:

  1. Stop defaulting to evaporative cooling
    • Switch to closed-loop systems and direct-to-chip cooling
    • Use air-side economization where possible
  2. Adopt water-free or water-lite technologies
    • Like Microsoft’s zero-water chip cooling
    • Or Lenovo’s Neptune™ direct recycling systems
  3. Let AI manage its own cooling
    • Google + DeepMind already use AI to reduce cooling costs by 30%
    • More companies should follow
  4. Stop building data centres in water-stressed regions
    • Prioritize cold climates and water-abundant regions
    • Avoid desert or drought-prone areas entirely
  5. Publicly report water usage
    • Not just emissions—we want litres
    • Break it down by product, region, and model
  6. Invest in water reclamation and protection
    • Fund wetland restoration, stormwater infrastructure, and Indigenous-led stewardship where they operate
  7. Heat Re-Use Projects
    • Innovations like IBM’s Aquasar reroute waste heat to warm buildings, further reducing water waste.
  8. Canadian Innovation: Deep Lake Water Cooling (Toronto)
    • This tech uses cold water from Lake Ontario to cool 100+ downtown buildings.
    • 75% less electricity
    • 40,000 tonnes of carbon dioxide eliminated annually

Until they do that, every new AI tool is just digital convenience built on environmental extraction.


Final Thoughts: If Feminism Doesn’t Include Climate Justice, It’s Just Branding

At She Zine Mag, we don’t see AI as the enemy—but we do see it as a responsibility. Every tool we use has a footprint. Every click means something.

This isn’t just about code or convenience. It’s about real water. Real ecosystems. Real damage we can either ignore—or repair.

We’ll keep making weird, defiant, joyful media.
We’ll keep experimenting with tech, tools, and storytelling.
But we’ll also keep asking:

What’s the true cost of this image, this edit, this template?
And how do we pay it back—with purpose?

→ Got tips, tools, or radical ideas? We want to hear them.
→ Want to help fund our next rain garden or tree planting? Let’s talk.

Because water isn’t just a metaphor. It’s the medium.
And if we’re gonna make a mess on this planet,
We’d better be ready to clean up after ourselves.

✷ COMING SOON ✷

  • Our Quarterly Reclamation Report
  • An interactive Prompt-to-Water Tracker
  • A merch line that offsets litres, not just emissions

This isn’t about perfection. It’s about practice.
Because reclamation isn’t charity—it’s resistance.

Subscribe to The Edit
Submit Your Work
Follow @shezinemagazine


sources: datacenterdynamics.com, smartwatermagazine.com, World Economic Forum, Googleblog.blogspot.com, livemint.com, CBS8.com, Wiki, nationalmagazine.ca, financialtimes.com, Wiki

Subscribe to The Edit for regular how-to's and special invites to exclusive workshops!!

By pressing the Subscribe button, you confirm that you have read and are agreeing to our Privacy and Terms of Use.

Add a comment Add a comment

Leave a Reply

Your email address will not be published. Required fields are marked *

Previous Post

Beyond the Ban: How Packaging Became the Frontline of Climate Justice

Next Post
A Mannequin wearing a craftivist mask

The Revolution Has Hot Glue On It's Fingertips: Why DIY Culture Is Political