Try Free

Study reveals energy demands and emissions of America's data centers

PBS NewsHour November 18, 2026 5m 945 words
▶ Watch original video

About this transcript: This is a full AI-generated transcript of Study reveals energy demands and emissions of America's data centers from PBS NewsHour, published November 18, 2026. The transcript contains 945 words with timestamps and was generated using Whisper AI.

"The 400 largest data centers in the United States emit more greenhouse gas emissions into the air in one year than the entire nation of Denmark. That's according to a new report from Harvard University. The study found that from May 2024 to April 2025, data centers accounted for nearly 2 percent of"

[00:00:00] Speaker 1: The 400 largest data centers in the United States emit more greenhouse gas emissions into the air in one year than the entire nation of Denmark. That's according to a new report from Harvard University. The study found that from May 2024 to April 2025, data centers accounted for nearly 2 percent of total U.S. energy consumption and produced more than 25 million metric tons of carbon dioxide, a level two or three times higher than an earlier estimate in 2018. I'm joined now by the lead author of that study, Gianluca Wiedi, a postdoctoral research fellow at Harvard focusing on AI and public health. Gianluca, welcome to the NewsHour. Thanks for joining us. [00:00:42] Gianluca Wiedi: Yeah, thank you for having me. Hello, everyone. [00:00:44] Speaker 1: So what was most surprising to you from this report? What really stood out to you in the way of the findings? [00:00:50] Gianluca Wiedi: Yeah, so think about it, right? So Denmark is an entire country and with millions of people, homes, hospitals and industries. And we're talking about like comparable annual carbon emissions from just 403 data centers. Those are just like 400 buildings. And that's an extraordinary scale. And those facilities are just only a fraction of the industry and how the industry looks like right now. [00:01:18] Speaker 1: You point out in the report, too, you weren't even able to capture all of the information from many data centers. So what do we not yet know about the impact that they could be having? [00:01:29] Gianluca Wiedi: Yeah, that's a great question. The study dates back like it's almost a year old by now. And since then, there's been like an incredible build out. We have preliminary results of newer studies that tells us there are around a thousand hyperscale data centers. So that's just more than the double of what we reported in our study. [00:01:53] Speaker 1: So many more data centers, meaning the numbers in here. In actuality now, the scope of emissions could be much higher than your report suggests. Is that correct? [00:02:02] Gianluca Wiedi: That is correct. Yeah, that is correct. And together with emissions, you have the water consumption and potentially also public health consequences. So in a way, we have that technology that is actually moving faster than our ability to understand its consequences. [00:02:18] Speaker 1: Tell us more about that. The intersection of the build out of these data centers and the impact on public health. What are you finding there? [00:02:27] Gianluca Wiedi: Yeah, in this specific study, we haven't addressed yet the public health. But what's happening basically is that once you have a power demand and a power that is produced at the power plant level, not only you're producing, we're producing carbon emissions, but we're also producing ads are those are pollutants that travel and eventually somebody breathes them. And the effects for for the health are pretty dangerous. And there's plenty and ton of scientific research about this. [00:02:59] Speaker 1: The power pull as well. When we talk about the power use in of these data centers, there are states you point out like Virginia, Iowa, Oregon and Ohio that account for half of the electricity use. Why is that? And what kind of downstream impacts does that have for residents in the area? [00:03:17] Gianluca Wiedi: Yeah, that's correct. So basically, the states offer the infrastructure, the land and electricity that large data centers need. But there's a catch. So the benefits of this expansion aren't necessarily felt by the same communities that bear its environmental costs. So more electricity demands can mean more power plants and therefore, as we're saying, more pollution and more pressure on local infrastructure. So those consequences don't stop at the data centers fence. [00:03:49] Speaker 1: There are those who will argue, John Lucas, I'm sure you've heard that we need to build out these AI data centers, that the country has to do it to remain competitive, that it's necessary for our economy, for national security, and that the impact of that on the environment or for residents in the area is a necessary evil. In other words, would you agree with that? Is that can that argument be made? [00:04:11] Gianluca Wiedi: Yeah, I see that. And there's a big backlash happening right now. I wouldn't call them like a necessary evil. I mean, I am an AI researcher myself, and I have to say, I'm really excited about this technology and what it can do, but there are better ways to build it. So they might cost more, they might take longer. The question is whether we were willing to prioritize doing it right over doing it first. [00:04:39] Speaker 1: The report, as you mentioned, is now several months old, almost a year old. There are a lot of things you don't address in it. For the average citizen to understand how this might impact their lives, what are the questions you think need to be answered next? [00:04:52] Gianluca Wiedi: Yeah, here's what concerns me the most. We're building like this AI infrastructure faster than we can independently understand its consequences. So our study, as you were saying, was just published, like, but the data are already like over a year old. So scientific evidence shouldn't arrive after the infrastructure and neither should, I believe, the public participation. And the main concerns would be the water consumption, the electricity, such that putting the power grid into, like, in strain and potentially, therefore, as I was saying before, the public health consequences. [00:05:32] Speaker 1: That is Gianluco Guidi of Harvard. He's a postdoctoral research fellow there focusing on AI and public health. Gianluco, thank you so much for your time. It's good to speak with you. [00:05:41] Gianluca Wiedi: Yeah, thank you so much for having me. Have a good night. Have a good night.

Transcribe Any Video or Podcast — Free

Paste a URL and get a full AI-powered transcript in minutes. Try ScribeHawk →