This is a podcast titled "AI's REAL Carbon Cost Might Surprise You" recorded in July 2025. It features a monologue by Jim Carter discussing the environmental impact of AI, including the carbon footprint of data centers and AI training, and highlighting industry efforts to improve sustainability. The episode also offers practical tips for listeners to reduce their AI-related carbon emissions and promotes Jim Carter's Slack community for AI enthusiasts.
Hey everyone. Welcome back to the Prompt, an AI powered podcast exploring what's possible with AI and my personal efforts in pushing the technology forward. I'm your host, Jim Carter, and today we're talking about something that's been on my mind a lot lately. AI's true carbon cost and what we're doing about it. You know, as much as I love the amazing potential of artificial intelligence, I can't ignore the fact that it comes with a pretty big environmental price tag. Training. Just one advanced AI model can use as much electricity as an entire neighborhood uses in a year. That's wild, right? But here's the thing. The AI industry is starting to wake up to this reality and they're making some big moves to address it. For starters, about 70% of new data centers are now connecting directly to wind or solar farms for their power. That's a huge shift to clean energy, and it's happening fast. Now, don't get me wrong, we've still got a long way to go. The carbon footprint of AI is just a small part of the bigger problem right now. But the important thing is we're on the right track and it's only going to get better over time. And let me give you a few examples of how companies are tackling this head on Google DeepMind came up with this smart idea to let an AI system control their data center cooling. It makes these tiny adjustments every five minutes and the result, they cut cooling costs by 40%. That's millions of dollars saved and countless tons of carbon that never entered our atmosphere. Then you've got Nvidia completely redesigning its chips. Their newest processors deliver twice the performance while using the same amount of power. And as models from just three years ago, it's like getting a car that goes twice as fast on the same tank of gas. And check this out. Smart startups like Anthropic are now timing their biggest training jobs for off times when wind farms generate extra clean power. Instead of pulling from coal plants during peak afternoon hours, they're riding the renewable energy waves. How cool is that? The common thread in all of this is that companies are realizing AI isn't just part of the problem, it's part of the solution. They're using the power of AI to make AI itself more sustainable. It's this beautiful virtue. Now, I asked my AI assistant to draw me a chart of what the end user carbon output looks like for average consumer tasks. Things like streaming Netflix, browsing social media, stuff we all do every day. And while it's definitely eye opening, it doesn't even begin to cover the emissions from heating, cooling, manufacturing, shipping, air travel. You get the idea. To put it in perspective, all US data centers combined produce around 106 million metric tons of CO2 annually. That's close to commercial aviation, which clocks in at 131 million metric tons. But here's the kicker. These are still only 1 to 2% of emissions in the global sector. So while it's a big number, it's important we keep it in context and stay informed on the facts. So what does all this mean for your daily life? Well, here's the best part. You're in control of what you use and what you care about. I like to find ways to have AI support being part of the change, not the other way around, like your voice assistant. It's not just answering questions, it's learning your habits to save energy. It might turn down the heat when you're out or lower the lights before bed. All powered by a chip that uses less energy than a night light. My kids use AI for homework. I use it to meal plan with my wife. If you're running projects, juggling deadlines, or managing a team, AI can help you do more with less fewer meetings, faster research, cleaner output. And when those tools are optimized to use less energy, you're not just saving time, you're lowering your footprint without lifting a finger. Once it runs on autopilot, it's a win win. So here are three things you can do this week to be part of the solution. First, time your compute tasks and cycle off when done. Second, reuse old gear instead of always buying new. And third, track your emissions and look for ways to reduce the future of sustainable AI is bright. New models are already 50% more efficient than those from just a few years ago. Open source tools are shrinking model sizes to run on laptops, opening up serious opportunities for schools, nonprofits and small businesses. These are just the start. But the good news is, while AI usage is skyrocketing, we're not suffering at the same rate with waste. It's all about balance. I'd love to hear your thoughts on this. Your real world experiences often shape what I cover in future episodes. If you're curious about the inner workings of this AI driven podcasting venture, I'm an open book. I'm building this in the public eye inside my Slack community. It's a safe space where we explore the strategies, scripts and workflows of AI in podcasting and so many other topics which I love to lead. If this kind of work lights you up. Come hang out in my new Slack community where we go deep on real world AI, use creative workflows and building smarter systems. I share everything I'm testing behind the scenes and it's become my favorite way to teach, collaborate and help. No gatekeeping, no fluff, and way more affordable than hiring me or my team one on one. Thanks for tuning in and let's keep pushing the limits of what's possible with AI1 prompt@ a time. See you in the next episode. Sam.
