[00:00:02] Speaker A: What if I told you that every time you ask ChatGPT a question, you're using 10 times more energy than a Google search? And the data centers powering AI, well, they could consume as much electricity as the entire country of Japan by 2030.
Here's why this matters.
The cloud, well, it isn't floating in the sky. No, it's burning through our planet's resources.
Cue the intro.
[00:00:40] Speaker B: From the Natural State. This is Public Facing, the podcast with Joe Holman.
[00:00:50] Speaker A: Welcome to Public Facing. I'm Joe Holman and we're here at part three of our AI and Innovation series.
Last episode we explored who controls AI. Today we're looking at the physical infrastructure that makes it all possible.
Data centers, the massive facilities housing the servers that power every AI interaction.
So here's observation that kind of blew my mind was Meta is actually planning to build a single data center the size of Central Paris.
That's not a typo.
Paris sized.
On today's roadmap, we're going to examine the explosive growth of AI energy consumption. We'll explore the environmental trade offs and consider what who wins and who loses in this digital infrastructure race.
Now, our main takeaway today would be understanding the true cost of AI infrastructure and that understanding that is essential for making informed decisions about our digital future.
All right, let's get into it.
Now.
With AI energy consumption growing so exponentially, the current state of affairs is that data centers currently consume about 460 terawatt hours globally.
That's roughly 1.3% of world world electricity.
Now, AI accounts for about 10 to 15% of that.
Here's some projection shock for you.
By 2030, data center consumption could reach 1,000 to even 2,000 terawatt hours.
Now, in the US alone, data centers could account for up to 13% of total electricity consumption. And that's up from 4% today.
Now, I want to give you the Virginia example.
Because one state, Virginia, now uses 25% of its electricity for data centers.
That too is not a misprint when it comes to Stargate. President Trump's Stargate.
At scale, the new $500 billion Stargate project will require approximately 25 gigawatts of uninterrupted power.
That's the equivalent to about 20 nuclear reactors.
Little fact here, a single ChatGPT query, it uses about 2.9 watt hours.
Now, that might not sound like much until you realize that billions of inquiries happen every day. It's like saying, I only eat one chip and then you multiply that by the entire Internet.
The digital world, it has a very physical footprint and AI is dramatically accelerating its growth.
Since we have covered the scale of things, I want us to take a look now at the environmental trade offs, the green dreams versus the gray reality.
Now big tech's climate commitments, they're colliding with AI's energy demands.
And there's broken promises out there.
Microsoft recently admitted its AI push has increased carbon emissions by 30%, jeopardizing its goal to be carbon negative by 2030.
And the water consumption, These data centers require vast amounts of water for cooling.
In the US alone, data center water usage equals a modern city, 3 million people.
And there is a sort of efficiency paradox here because while chip efficiency improves dramatically, 100,000 times improvement over 10 years, well the rebound effect, well that means that we just use more AI, not less energy overall.
And let me point out the nuclear renaissance here because Amazon, Microsoft and Google, they're signing deals with nuclear operators.
Microsoft even reopened the Three Mile island plant to power its AI operations.
Microsoft's president essentially said that good AI will outweigh, should say quote, will outweigh its environmental impact, unquote.
That's certainly one way to look at it. Another way might be, quote, sorry about the planet, but check out this really good chatbot.
I think there is a trade off there.
Now the sustainability of AI isn't just a technical problem.
It's a fundamental question about what kind of future we're building.
And there's a whole other side to this because who pays the price exactly?
The costs and the benefits of AI infrastructure, well they aren't evenly distributed really, not by a long shot.
There are concentrated benefits such as an estimated 84% of AI's projected future $15.7 trillion economic value will flow to China, North America and Europe.
But there are local burdens such as where Virginia, they gave up $750 million in tax revenue through data center incentives.
And that was just in 2023 alone.
The state of Illinois spent $468 million in subsidies, creating just 339 jobs, about $1.4 million per position.
Now we're feeling the effects of the energy price where Ireland's wholesale power prices, they're a third higher than than the European average. And this is driven partly by data centers consuming over 20% of their national electricity.
And there is a bit of a deep SEQ disruption. What do I mean? Well, Chinese AI company, Deep Seq, they claim to have trained comparable models using far less energy and cost that suggest that more efficient paths may be possible.
But let's talk about sustainability conversations.
The 2025 Paris AI Action Summit emphasized sustainable AI, with new initiatives launching to track and reduce environmental impact.
My thyl thought the data center boom is real and it's accelerating. And it's got its consequences environmental, economic, and even social.
But these are only beginning to be understood now. Every AI conversation, every generated image, every automated decision, well, it has a physical cost somewhere on the planet.
In our next episode, we're going to put safety first by examining the real risks of engaging with AI, from deep fakes to power displacement. Because knowledge isn't just power, it's protection.
And that's going to bring us to today's final Word.
In Today's final word, AI's energy consumption is growing exponentially, with data centers potentially reaching Japan level electricity use by 2030.
Now, Big Tech's climate commitments, they're struggling against AI's insatiable appetite for power.
The costs, environmental, financial and social aren't evenly shared across communities or nations. Nowhere close.
But ultimately, the infrastructure powering our digital future is a choice that we're making collectively, whether we realize it or not.
And that's today's final word.
If you enjoyed today's episode, please make sure to follow and share Public Facing wherever you listen.
Email us direct helloblicfacingpod Stream.
Visit us at Public facingpod Stream for all of our episodes and that's going to do it for our time together today. If this episode has opened your eyes to the hidden costs of AI, please do me a favor. Share this with a friend or even a colleague and make sure you follow and stay connected to Public Facing the podcast for more conversations about the people, issues and everyday moments shaping our world.
Thanks for spending time with us today for Public Facing, I'm Joe Holman. Come back and join us again real soon.
Until next time, stay informed, stay encouraged, and stay Public Facing.
[00:11:32] Speaker B: Thanks for listening to Public Facing, the podcast with Joe Holman. Visit us online at public facingpod stream and on
[email protected] publicfacingpod Sam.
