How much energy and water does ChatGPT consume?
If you purchase an independently reviewed product or service through a link on our website, BGR may receive an affiliate commission.
Artificial intelligence has been the hottest topic in tech since late 2022, when ChatGPT went viral. The AI race started almost immediately, with every big tech company in the US and elsewhere working on new AI systems of their own.
We quickly learned that software like ChatGPT requires massive resources. Datacenters packing thousands of expensive GPUs specialized in training and running AI chatbots were needed. The larger the data centers, the more energy the world would need to set aside for AI projects.
Today's Top Deals
Best deals: Tech, laptops, TVs, and more sales
Best Ring Video Doorbell deals
Memorial Day security camera deals: Reolink's unbeatable sale has prices from $29.98
Some people worried about the impact AI infrastructure would have on the world. It wasn't just about the electricity powering the chats, but also the water used to cool some of these data centers.
Two and a half years after ChatGPT went viral, we finally know how much energy and power a ChatGPT chat consumes. It comes from Sam Altman's latest blog, titled The Gentle Singularity, which teases what the world could look like in the next five to ten years thanks to superintelligence:
People are often curious about how much energy a ChatGPT query uses; the average query uses about 0.34 watt-hours, about what an oven would use in a little over one second, or a high-efficiency lightbulb would use in a couple of minutes. It also uses about 0.000085 gallons of water; roughly one fifteenth of a teaspoon.
It's unclear where Altman's figure comes from. If accurate, I'll probably consume about a teaspoon of water with AI queries every day. But the figure is also misleading, considering that OpenAI has hundreds of millions of monthly users.
Add to that all the energy and water Gemini, Claude, Meta AI, Deep Research, and all the other chatbots out there consume, and you'll rack up quite a bill.
It's unclear what prompted the CEO to pen the post, one he describes on X as maybe 'the last one like this I write with no AI help at all.' It likely wasn't to reveal the energy costs associated with ChatGPT chats, though energy is one of the big topics in the blog. On that note, I'll point out that ChatGPT o3 just got dramatically cheaper than before.
Unsurprisingly, Altman is quite optimistic about the future of AI. He presents superintelligence as inevitable. It's a foregone conclusion. We'll get to a world where smarter-than-human AI will make our jobs easier than ever, leading to potential massive discoveries to improve daily lives:
AI will contribute to the world in many ways, but the gains to quality of life from AI driving faster scientific progress and increased productivity will be enormous; the future can be vastly better than the present. Scientific progress is the biggest driver of overall progress; it's hugely exciting to think about how much more we could have.
Altman expects AI to bring novel insights in 2026. A year after that, the world will start getting robots that can do tasks in the wild. Then, 'the 2030s are likely going to be wildly different from any time that has come before,' Altman writes. That's even though we, humans, will continue to enjoy our lives as we did before.
The OpenAI CEO also says that intelligence and energy will be 'wildly abundant' in the 2030s. Once that happens, the world will be able to do things that weren't possible before.
Speaking of energy, Altman also sees datacenter production becoming automated. AI and robots will power everything:
There are other self-reinforcing loops at play. The economic value creation has started a flywheel of compounding infrastructure buildout to run these increasingly-powerful AI systems. And robots that can build other robots (and in some sense, datacenters that can build other datacenters) aren't that far off.
AI will help humanity achieve new 'wonders' by 2035. Altman even sees a future where some people will choose to 'plug in' via 'true high-bandwidth brain-computer interfaces.'
It all sounds amazing, and it certainly beats the gloomier pictures others paint about the future of AI.
Altman's essay also downplays the downsides, like the massive job revolution we're about to witness. The CEO isn't ready to propose any solution for AI stealing jobs, other than indicating that humans will adapt and some sort of new social contract might emerge:
There will be very hard parts like whole classes of jobs going away, but on the other hand, the world will be getting so much richer so quickly that we'll be able to seriously entertain new policy ideas we never could before. We probably won't adopt a new social contract all at once, but when we look back in a few decades, the gradual changes will have amounted to something big.
Toward the end of the post, Altman also addresses the obvious challenges. AI has to be aligned to our interests to give us the rosy future he paints in the previous paragraphs. The other challenge is making sure superintelligence is 'cheap, widely available, and not too concentrated with any person, company, or country.'
How will OpenAI and every other firm engaged in developing frontier AI ensure it's safe, cheap, and widely available? Altman doesn't say. We'll just have to wait and see what happens next, while we continue to chat with chatbots like ChatGPT, a fifteenth of a teaspoon of water at a time.
Don't Miss: Today's deals: Nintendo Switch games, $5 smart plugs, $150 Vizio soundbar, $100 Beats Pill speaker, more
More Top Deals
Amazon gift card deals, offers & coupons 2025: Get $2,000+ free
See the
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Yahoo
19 minutes ago
- Yahoo
ChatGPT may be the smartest software ever, but this Pong-era game console can do something it can't
When you buy through links on our articles, Future and its syndication partners may earn a commission. Technology. The ever-advancing pinnacle of the processor. It's a modern-day blessing to most, but dreaded by those convinced it's five minutes away from flipping society onto its head like an egg served sunny-side down. The word alone can strike fear into the hearts of man. Especially if you shout it loud enough into their ears as you pass them on the street. And, if you were to take a small break from harassing the general public and ask one of them to name today's most advanced piece of technology, they'd likely say ChatGPT — OpenAI's super-brainy chatbot, packed with enough artificial intelligence to seemingly make regular intelligence look like its eating glue from a pot in a sandbox. But if ChatGPT, the poster child for cutting-edge technology, is so smart, how did it just get absolutely bodied by a video game console released in 1977? When we think of competitors to ChatGPT, there's a usual list of suspects to choose from: Google Gemini, Microsoft Copilot, Meta AI, Anthropic's Claude, Perplexity AI, and maybe even newcomer DeepSeek. What you wouldn't expect to appear in that list is the Atari 2600, a 48-year-old home video game console best known for bringing Pac-Man into the living rooms of millions of first-generation gamers. However, thanks to Citrix Engineer Robert Jr. Caruso, Atari's retro console can now be counted among ChatGPT's truest rivals, after it was used to repeatedly best OpenAI's GPT-4o model at a simple game of chess. In a now-viral post shared to LinkedIn, Caruso details his 90-minute experiment in pitting the computing might of tens of thousands of Nvidia GPUs against the singular 1.19 MHz 8-bit MOS Technology 6507 processor of the Atari 2600, claiming ChatGPT "made enough blunders to get laughed out of a 3rd grade chess club." Clearly, they don't make them like they used to. The prevalence of the 2600 over ChatGPT is a true David vs. Goliath battle on a checkerboard stage, and a surprising outcome to most. However, there's a chasm of difference between chatbot and a chess engine, meaning the Atari 2600, which ran not as hardware but through the Stella emulator, likely had ChatGPT's number from the start. While OpenAI has made great strides in improving its model's memory capabilities, it's still primarily a language prediction machine, and not the next Deep Blue. Still, given the 2600's ability to only predict two moves in advance, it does highlight ChatGPT's shortcomings, and provide a thumb to the virtual eye for OpenAI's world's most intelligent chatbot. At least it would do if OpenAI had ever claimed as much. In fact, it's more often than not proclaiming the opposite. Stretching back to 2023, in an episode of the Lex Friedman podcast, OpenAI CEO Sam Altman was quick to label GPT-4 as "a very early AI. It's slow, it is buggy, and it does not do a lot of things very well." Then, in 2024, during a Q&A at Stanford University, Altman claimed that ChatGPT was running on "the dumbest model any of you will ever have to use again by a lot." In fact, only recently has Altman pushed the message that AI is living up to its supposed smarts. In a blog post published on Tuesday, titled The Gentle Singularity, Altman predicts: "We do not know how far beyond human-level intelligence we can go, but we are about to find out." So yes, Atari's classic console may have bested ChatGPT this time, but if Altman's words are anything to go by, it could be a very short-lived victory. Sam Altman's vision is the stuff of my nightmares: "A significant fraction of the power on Earth should be spent running AI" Wikipedia editors revolt over "truly ghastly" plan for AI slop — they're winning (for now) Microsoft has created an after-the-fact AI gaffer -- but is your laptop up to snuff?
Yahoo
41 minutes ago
- Yahoo
What Nvidia's Jensen Huang and OpenAI's Sam Altman Had to Say This Week About AI and Jobs
Nvidia's Jensen Huang said this week he believes "everybody's jobs will be changed" by artificial intelligence. Also this week, OpenAI's Sam Altman said there could be "whole classes of jobs going away," due to AI, but that he expects people to find new solutions to address the shift. The comments come amid concerns over the extent to which AI could disrupt the labor (NVDA) CEO Jensen Huang said at an event earlier this week that he believes 'everybody's jobs will be changed" by artificial intelligence. "Some jobs will be obsolete, but many jobs are going to be created … Whenever companies are more productive, they hire more people," Huang said at the VivaTech conference in Paris Wednesday, Fortune reported. The comments were in part in response to remarks from Anthropic CEO Dario Amodei, who last month said that he expects AI could wipe out half of all entry-level office jobs, the report said. Huang said he 'pretty much disagree[d] with almost everything' Amodei said. He wasn't the only tech CEO who had something to say about AI and jobs this week, amid worries the rapidly developing technology could lead to significant job losses. Sam Altman, the CEO of ChatGPT maker OpenAI and one of the faces of the AI boom, wrote in a blog post Wednesday that he expects there could be "whole classes of jobs going away" as AI develops. ChatGPT "is already more powerful than any human who has ever lived,' said Altman, who sees a future in which robots could be capable of building other robots designed for tasks in the physical world. However, Altman added "people are capable of adapting to almost anything," and said he expects humanity to find new solutions to address the shift. Read the original article on Investopedia Error in retrieving data Sign in to access your portfolio Error in retrieving data Error in retrieving data Error in retrieving data Error in retrieving data
CNET
an hour ago
- CNET
'Textbook Copyright Infringement': Disney and Universal Sue AI Image Generator Midjourney
Disney, Universal and several of their entertainment companies filed a lawsuit against popular AI creative service Midjourney on Wednesday, alleging that the company committed copyright infringement. It's a big move from power players and will no doubt create ripple effects across the AI and entertainment industries that will flow all the way to what you can create using AI tools. Midjourney is one of many AI image generators that use generative AI text-to-image technology. With an account, anyone can use its models to create digital images. Many AI image generators have policies and internal guardrails that prevent people from recreating brand logos, celebrity likenesses and other kinds of recognizable and sometimes copyrighted material. Disney and Universal allege that Midjourney did not take these precautions even after they reached out to the company to express their concerns. The companies said in the lawsuit that Midjourney's AI image- and upcoming video-generation technologies "blatantly incorporate and copy Disney's and Universal's famous characters" without proper licensing or having a hand in original creation. "Midjourney is the quintessential copyright free-rider and a bottomless pit of plagiarism," they wrote. Midjourney did not respond to a request for comment. The 100-plus-page lawsuit details the ways Midjourney enables its users to recreate characters that belong to Disney's and Universal's different worlds, such as Marvel and Star Wars. It includes examples of images the companies were able to generate using Midjourney that feature some of their iconic characters, including those from Shrek, Star Wars and How to Train Your Dragon. Disney included these images in its lawsuit to show AI images created with Midjourney that mimic copyrighted characters. Screenshot by CNET Copyright is one of the core legal and ethical issues in AI, and this is far from the first lawsuit between entertainment companies and AI companies. There is an ongoing class-action lawsuit from a collection of artists, led by Karla Ortiz, against Stability AI. Publishers like the New York Times are also concerned, suing ChatGPT maker OpenAI. (Disclosure: Ziff Davis, CNET's parent company, in April filed a lawsuit against OpenAI, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.) At the same time, some entertainment companies are slowly exploring ways to integrate AI into their creative workflows. Disney has been fairly mum about AI, neither endorsing or seeking partnerships like its entertainment peers at Lionsgate, but not publicly ruling out the possibility either. That possibility is reflected in the statement Disney made to CNET via email. "We are bullish on the promise of AI technology and optimistic about how it can be used responsibly as a tool to further human creativity," Horacio Gutierrez, senior executive vice president and chief legal and compliance officer, said in the statement. "But piracy is piracy, and the fact that it's done by an AI company does not make it any less infringing." Another example Disney cites in its lawsuit. Screenshot by CNET Read more: Inside Hollywood's AI Power Struggle: Where Does Human Creativity Go From Here? The lawsuit marks a path forward for Disney and Universal and adds another strand to an already tangled legal web. "The lawsuit filed by Disney and Universal is important in drawing a line in the sand with AI developers like Midjourney," Robert Rosenberg, an intellectual property lawyer and former general counsel at Showtime Networks, wrote in an email. "As the lawsuit explains, the only way the AI platforms can output an image of Yoda, Shrek or Darth Vader is because they have trained their model by ingesting copyrighted images of these characters. They are not inventing new characters here." For now, we'll have to wait and see how this case and the other court cases progress. In the meantime, Midjourney users and other AI users are able to continue utilizing those services. For more, check out our guide to understanding copyright in the age of AI.
