Feeding the Matrix: a special series on AI and climate

To kick off 2025, Everybody in the Pool is talking to the giants of AI about how to make data centers more efficient, and innovators in the field about how to reinvent them.

Welcome back to Everybody in the Pool!

It’s obviously a wild start to 2025, and I’m going to acknowledge up front that the landscape has changed dramatically, there’s a lot of anxiety and tension in the air around what’s going to happen with climate policy in a time of well, slashing and burning anything that looks like climate policy.

Here’s my shortest possible take: the work goes on. The work of innovators, inventors, scientists, evangelists, activists, entertainers, and investors will never be more important than it is right now. I often tell people that I believe in science — and that includes the Newtonian principle that every action has an equal and opposite reaction. If you are feeling afraid about what the future holds, if you’re feeling like our ability to address this crisis is either slipping or being yanked violently away from us, then I am here to urge you to be the equal and the opposite. I still believe that together, we can get this done.

So let’s get to what’s happening on Everybody in the Pool, shall we? Toward the end of last year, I realized my life as a tech reporter and my life as a climate tech reporter were on a collision course, thanks to AI. The arrival and subsequent investment in more and more and more computing power means a corresponding increase in energy consumption, a rush to construct more data centers, and a related fear of what might happen to global emissions as a result. So I’m kicking off 2025 with a four-part series on AI, data centers, and climate change that I’m calling “Feeding the Matrix.”

First, some numbers:

• In 2023, according to Lawrence Berkeley National Laboratory, data centers consumed 4.4% of US electricity.

• That number, per LBL, could reach 12% of US electricity by 2028, more than the electricity used by California, Florida, and New Jersey combined.

• Amazon, Google, Meta and other big tech companies have all announced investments ranging from $80 billion to $150 billion in new data center construction

• OpenAI, Softbank, and Oracle recently announced (at a White House press conference, oddly enough) plans to spend $100 billion on 10 data centers in the US over the next 10 years, with potential to scale to $500 billion in investment and 20 or more total facilities (of 500,000 square feet each)

• Big tech companies are desperate for power, making investments in fusion energy, small nuclear reactors, existing nuclear reactors, solar and wind, and in the case of Stargate, OH YAY, a big new fossil fuel power plant backed up by diesel generators.

You can kind of see how the first couple of bullet points, combined with the later bullet points, might lead to a fairly obvious conclusion: Yikes.

Shout-out to Hewlett Packard Labs, who tried to sound the alarm back in 2015. Ha. Haha. Aw.

A now creative-commons image hoping to alert us that data centers might eat us all eventually. They tried!

In fact, some of you may remember I wrote about this back in 2023, and I cited the great American prophecy, The Matrix, in wondering how soon it might be before we scorch the earth and start using humans for batteries.

So what are we talking about in this series? Well, how to avoid that scenario, basically.

What We've Learned So Far

I've already had two fantastic conversations with massive companies at the absolute heart of this conversation. First up was Josh Parker, who heads up corporate sustainability at Nvidia. They're the undisputed champions of AI chips and software (notwithstanding, um, DeepSeek), and Josh walked me through how they're making their chip architecture more efficient with every iteration — which is crucial since practically every AI model out there runs on their tech (notwithstanding, um, DeepSeek).

Josh is also hopeful that AI itself will help solve some of AI’s energy drawbacks. For example, the ability to make highly complex digital twins of new data centers or manufacturing plants helps design them much more efficiently in the real world — something Nvidia’s partners are already doing.

Then this week, I sat down with Chris Walker from AWS (aka Amazon's cloud empire, and the biggest player in the game by market share). He made a compelling case that AI won't derail companies' sustainability goals, partly because their customers are demanding energy-efficient data centers – both for the planet and their bottom line. He also talked about how, even before innovative new data centers come online, there’s lots to be done with existing infrastructure, from AWS building its own custom chips to literally rearranging racks and servers so fiber and even vehicles have less energy-consuming ground to traverse. Think of it like swapping all the light bulbs for LED bulbs (they’re doing that, too). It ain’t sexy, but it cuts the electricity bill.

Coming Up Next: The Future of Data Centers

Now, here's where things get really exciting. In the second half of our series, we're looking at data centers of the future that could be energy independent, maybe even net positive, and actually welcome in communities instead of being seen as resource-hungry monsters.

Next week, I'm sitting down with Kieran Furlong, the co-founder of Realta Fusion. It’s a fusion energy company developing reactors that are compact enough to build right next to data centers. They’d create little islands of clean, abundant energy that don't strain the surrounding power grid at all — hopefully clearing the way for much faster construction and getting future data centers off of public infrastructure at all.

And for our series finale (for now, anyway), I talked with Joann Garber about her fascinating work at Microsoft on what they call "regenerative" data centers. She completely flipped the script on how we think about computing infrastructure, reimagining these facilities as something that could actually give back to landscapes and communities instead of depleting them.

What's at Stake

The decisions we make now about AI infrastructure will have lasting implications for global sustainability goals. If we get this wrong, the AI boom could derail climate targets and accelerate environmental degradation. But if we get it right, AI could become not just energy-efficient but a powerful tool in our fight against climate change.

Stay tuned for these upcoming episodes – I genuinely think they're some of the most important conversations we've had on the show.

Subscribe now to ensure you don't miss any part of this crucial series. And as always, thanks for listening!