Navigating the Data Center: The Journey of Your AI Prompt
In a world increasingly reliant on artificial intelligence, understanding the energy implications of our digital interactions is crucial. WSJ’s latest video, "How Your AI Prompt Travels Through a Data Center," dives into the intricacies of this unseen journey, revealing how AI could consume up to 12% of all U.S. electricity by 2028.
From Request to Response
The video begins in “Data Center Valley” in Virginia, where WSJ’s Joanna Stern takes viewers on an enlightening journey. After you hit “enter” on that AI prompt, your request doesn’t just vanish into thin air. Instead, it sets off a complex series of events through a vast digital landscape. Stern meticulously traces how data travels through layers of infrastructure before arriving back to you.
Energy Usage Breakdown
The highlights of this exploration come as Stern addresses perhaps the most pressing concern—energy consumption. How much power does it truly take to create a single video or respond to an AI query? Audiences are presented with surprising statistics and visual representations that paint a clearer picture of what goes into our digital lives.
The Role of Nvidia GPUs
Central to this narrative is the role of Nvidia GPUs, the powerhouse behind many AI applications. At 1:34 into the video, viewers learn about the high energy consumption of these units and their significance in processing AI requests. Stern does not shy away from discussing the environmental footprint of the technology we often take for granted.
A Unique Analogy
In a clever twist, Stern uses her experience grilling steaks as a metaphor for energy consumption. By linking the preparation of a meal to the energy required for data processing, she effectively illustrates the often-overlooked parallels between our everyday activities and our tech-driven world.
Takeaways for the Future
As the video closes at 6:02, Stern leaves viewers with essential takeaways about the implications of AI on energy consumption. Advocating for awareness and greener practices, she encourages a critical examination of our digital habits and their broader impact on the planet.
With a compelling narrative and a blend of humor and insight, Joanna Stern delivers a timely exploration of technology that prompts both fascination and reflection. As society stands on the precipice of an AI-driven future, understanding these dynamics may become increasingly vital.
Watch the video by The Wall Street Journal
Video “How Your AI Prompt Travels Through a Data Center | WSJ” was uploaded on 07/01/2025 to Youtube Channel The Wall Street Journal
Yeah right nvidia
Based on their math…. a 6-second video costs about 1.5 pennies (0.1kwh with a kwh at $0.15). That is not alarming… I find more alarming that a single EV charge at home will be equivalent to a 4-hour movie being made by an AI, or using that little grill to make steaks for 12,000 people. This is only for 1 solo EV at one residential home.
I hope they use even more energy! Accelerate
I don't use AI-based suggestions on a daily basis, but companies like Google do AI-based summaries for my search queries without my asking, dozens of them every day. So they should be blamed, not the consumer, it's hard to imagine how many AI results are thrown in the trash, generated just to promote these hallucinations.
Misleading title! It's the energy of cooking a steak not a steak.
Awesome video, what an insane places. Data centers really are getting bigger and bigger share of our power,
I just hate that companies want to make the consumers responsible for the global warming impact, when they HAVE THE POWER to implement environmental policies that protect our planet.
Gobbling up all of the graphics chips and consuming as much electricity as an entire state, all for a gimmick that can't get basic details correct.
I have exactly 2 big reasons only to watch this video.
He probably means 90 degrees celsius
and yet, people still think this is the next evolution step. No wonder we are doomed. You didn't discover anything new – all of that is known and easy to calculate or find out. What's troubling is the ignorance of the people towards this massive waste.
It's remarkable how people can burn through 300-500 watts binge-watching Netflix documentaries about AI destroying the planet, while that same energy could have powered 1000+ ChatGPT queries to actually solve their problems. Apparently entertainment watts are morally superior to productivity watts.
I'm sorry, but shaming consumers for the power they use when using AI is not much different than shaming them for using plastic straws. Stop focusing on individuals and focus more on the massive billion dollar companies creating these wasteful and polluting products. Go lobby the government to regulate these industries, stop shaming and blaming people who are just trying to live their lives.
Honestly, I was surprised at how efficient AI generation is!
What a weird comparison. The vast majority of the GHG emissions for a steak are before you buy it, including growing feed, and methane from cattle. Total, it's 100-155 lb CO2 eq per lb of beef.
I wish this video has discussed the energy used in training, which provides the bulk of power consumption during the end-to-end process for generative AI. There are a lot of misleading statistics out there claiming that a single query consumes much more power or water than is claimed in the video, largely because they factor in energy consumed during training. It would have been interesting to discuss this side of things and how rising aggregate demand for generative AI could affect how much training is performed for newer frontier models.
Please give a link to the original research that you cite, rather than a link to your own pay-walled article on the topic.
A new Marvel supervillain will create a billion simultaneous prompts and cause a global meltdown.
what happens? wsj gets your data and money there and global warming, pollution, electricity bill rises
She just wasted energy and the resource.
Couldn't we just make them in places further up north like Iceland where it's colder?
You are an excellent technology columnist
so inefficient
Bro since when the wsj cared about climate change 😂😂
she's gonna have her own TV show or something like that
why do reporters have to use all those "how long these power would power the grill" analogy? also whenever something big gets mentioned, it's always equivalent to "x football fields". are american not able to comprehend basic concepts of physics or what?
Expecting people to hold off on AI usage to save energy is actual lunacy.
Where's the nuclear?
Do the AI on the inter-NETS
Very nice job. I have been been in the data center industry a while (maybe too long?!), and I like your approach to show and explain things. Very very well done. I share videos like this to new team members to help them wrap their head around what's going on.
Every pump jack in America uses more electricity than all the EVs in the US. How about we make fracking operations use Stirling engines to power those pads with waste methane they are flaring into the atmosphere. Lease oil land belongs to us and they're wasting the resources and polluting the skies to produce oil.
You are doing some dis-service by comparing energy usage of AI to cooking steak.
These data centers run the whole internet, not just AI
So asking ChatGBT to calculate PI to infinity is a bad idea.. got it.
Good information, but it’s not good to eat red meat. Most likely source of Alzheimer’s.
how many decades, years, before we're in the plot of the Animatrix with AI harvesting humans for bio electricity ??
how many decades, years, before we're in the plot of the Animatrix with AI harvesting humans for bio electricity ??
Yeah. People will definitely stop silly prompts after watching this✋
I mean why the models would not be even free for making these unnecessary videos, what's the use?
Yeah right…the end by the dude, "you know research on molecule that could save lives…" Let's depart with a clean conscience, shall we?!
Something that would be helpful, maybe in a future video, is to explain what is physically happening when these data centers are at work. What are the physical requirements that result in the energy use and why is so much of it turned into waste heat?
Read more about how much energy AI prompts use here: https://bit.ly/4esbmrk