The truth about AI’s carbon footprint

This Week

People ask me all the time:

“Isn’t AI bad for the environment?”

And up until now, I never had a solid answer. So this week, I dug in to see what’s really going on.

What I found was more complex (and more hopeful) than I expected. Yes, AI uses energy, but so does everything else we count on, from hot showers to sending emails. The important thing is understanding what that impact looks like and what we can do about it.

Here’s what you’ll find in this week’s edition:

• How Much Energy Does AI Really Use?

• Can AI Actually Help the Climate?

• A Mini Energy Audit for Knowledge Workers

• Six Questions to Ask Your AI Vendors About Sustainability

Curious how your AI use stacks up, and what you can do to make it cleaner? 

Read on!

Is AI a Power Hog? Not Compared to Netflix

There’s been a lot of mystery around how much energy AI really consumes. Partly because the biggest tech companies aren’t exactly rushing to hand over the data. But with pressure mounting, especially from industries like travel where sustainability is front and center, the energy debate around AI is heating up (see what I did there :D).

AI runs on electricity, just like any other digital tool, but its energy use spans a whole supply chain. From the power plant to the data center to your device, every prompt taps into a system that pulls power every step of the way.

So yes, AI uses energy but maybe not as much as you think.

Right now, global data center emissions are estimated at around 0.5%, compared to 2.5% from the aviation industry. But that gap is closing fast. As AI adoption surges, emissions from data centers are projected to hit 3.5% — outpacing aviation entirely.

The place you’re more likely to feel AI’s impact right now is in your energy bill. Across the U.S., households are seeing noticeable increases, with some regions in the Northeast seeing rates jump over 13%. That’s largely because data centers (which power and train AI models) are guzzling electricity, forcing utilities to upgrade the grid. And those costs are being passed straight onto you.

Perhaps a more pressing concern is data center water usage for cooling. In some locations this usage it putting serious strain on water supplies and only a few of the tech giants have pledged water sustainability goals.

Despite these impacts, I’d never tell anyone to skip using AI because of the energy cost. That’s like swearing off your laptop to save a few watts. Would you give up your morning coffee or your hot shower? All of those things have a higher energy “cost” than your prompt in ChatGPT.

According to Google’s latest carbon impact report, a typical text query burns through about 0.24 watt-hours of electricity — about as much as watching 9 seconds of Netflix. 

The better approach is figuring out how to build sustainability into your AI use.

Here are a few ideas:

• Check the Carbon Score of the tools you’re using

• See whether your model provider is investing in renewable energy (or not)

• Switch to a more energy-efficient laptop

• Use cloud services with green certifications

• Choose smaller models when possible

• Turn off auto-sync and background AI tools so they don’t vampire energy

• Run intensive tasks during off-peak hours

So yes, AI uses energy. But thanks to increasing transparency and smarter engineering, we’re getting better at managing the impact.

What If AI Is Actually Good For The Planet?

What if the very thing draining energy from the grid could be the key to fixing it?

That’s the argument from Crusoe President Cully Cavness, who believes AI might be one of our most powerful tools in the fight against climate change. He talks about how AI is helping to accelerate the shift toward clean, low-carbon energy.

The problem? Clean energy development is too often stuck in permitting purgatory. But AI can help cut through the red tape and speed up progress.

Here’s where it’s already making an impact:

• Streamlining clean energy permitting and project planning

• Improving supply and demand forecasting to stabilize the grid

• Enhancing energy storage and load balancing, from national grids to smart homes

• Accelerating the discovery of new battery materials and cleaner fuels

• Powering rapid design and testing of breakthrough tech like compact fusion

• Supporting more efficient scheduling of electricity production and use

For example, SES AI used AI to shrink a 8,000-year research task down to two months — leading to the world’s first batteries using AI-discovered electrolyte materials.

Companies like Avalanche Fusion are relying on AI to bring advanced, compact reactors closer to commercial use. Crusoe itself is building data centers that run on flared gas and stranded renewables, aiming to reduce the emissions impact of AI infrastructure at the source.

Cavness doesn’t deny that AI uses energy. But the bigger picture is this: the climate tech enabled by AI may end up saving more emissions than AI produces. The faster we harness that potential (and power it cleanly) the better shot we have at staying below catastrophic warming thresholds.

Read the full article at Fortune

What to Ask Your AI Vendors About Sustainability

If you're evaluating AI tools or platforms it's smart to ask a few questions about how they're handling energy use. Most vendors won’t bring it up unless you ask, but it’s quickly becoming a key part of responsible procurement.

Here are the questions I include in every vendor review:

☐ Do you publish the energy consumption or Carbon Score of your models? 
If not, why? Transparency here is a good indicator of whether they’re paying attention to their impact.

☐ Are your data centers powered by renewable energy? 
Look for vendors who either operate their own green data centers or use cloud providers with strong sustainability programs.

☐ What steps are you taking to reduce the environmental impact of AI training and inference? 
Training foundation models can be resource-intensive. A forward-thinking provider should be optimizing for efficiency.

☐ Are you focused on carbon offsets, or actual emissions reduction? 
Many companies rely on offsets to meet sustainability goals, but long-term solutions need to go beyond credits.

☐ Can your product run locally or on more efficient hardware?
Cloud-based tools aren't always the most sustainable. Vendors offering lightweight, on-device options can reduce both emissions and cost.

☐ Do you provide sustainability documentation? 
Especially useful for teams working under ESG guidelines or procurement requirements.

What’s Your AI Energy Footprint?

AI is now baked into most workflows — from writing emails to generating decks to storing everything in the cloud. Each step has a hidden energy cost, but there are small changes you can make to keep things lean.

Here’s a quick audit to help you spot where the energy’s going and how to use it more efficiently.

File Storage & Hosting

Tools: Google Drive, Dropbox, Notion, Canva, cloud asset managers
Energy use: Ongoing and often overlooked

Make it more efficient:

• Delete unused files and drafts regularly

• Turn off auto-backup features you don’t need

• Compress large media files before uploading

Writing Emails

Tools: Gmail AI, Superhuman, Outlook Copilot
Energy use: Low — but constant background assistance adds up 

Make it more efficient:

• Use AI writing help selectively, not for every sentence

• Disable predictive text or live suggestions if not useful

• Batch email writing to reduce idle tool usage

Creating Presentation Decks

Tools: Gamma, Tome, Canva, PowerPoint
Energy use: Medium — design automation can be compute-heavy

Make it more efficient:

• Build from templates instead of starting from scratch each time

• Limit real-time collaboration if it auto-syncs constantly

• Export only what you need, avoid redundant versions

Writing & Content Generation

Tools: ChatGPT, Claude, Jasper, WRITER
Energy use: Low per prompt — but frequent use adds up

Make it more efficient:

• Use smaller models for everyday tasks

• Batch your writing sessions to avoid idle compute time

Video Generation & Editing

Tools: Runway, Pika, Descript, Synthesia
Energy use: High — especially for rendering and generation
Make it more efficient:

• Render in lower resolutions where possible

• Avoid leaving tools open in the background

Image Generation

Tools: Midjourney, DALL·E, Firefly
Energy use: Medium to high depending on prompt complexity

Make it more efficient:

• Limit iterations, each prompt uses compute

• Reuse and adapt assets when possible

Automation & AI Scheduling

Tools: Zapier, Co-pilot assistants, AI plugins
Energy use: Low, but persistent

Make it more efficient:

• Turn off always-on workflows

• Use scheduled or manual triggers instead of live syncing

You don’t have to cut out all your tools or workflows. A few tweaks to your stack can reduce energy use without changing how you work, and you can feel good about your energy consumption. If you want to go further, look into which platforms use renewables, and check whether your favorite tools disclose their energy stats or carbon scores.

Want to Level Up Your AI Game?

If your team is ready for a hands-on AI strategy session, my custom-designed workshops are built to uncover the workflows that can save you hours every week.

Prefer to start small? My YouTube channel is packed with quick, practical “how-to” videos that show you exactly how I use AI tools for marketing, content, and automation.

Planning an event or conference? I deliver high-energy AI sessions that engage audiences and leave them with actionable strategies they’ll talk about long after the event. Book me for your event here.