The contribution of electricity consumption of GPT-5: As high as from 1.5 million households-first appeared at the online magazine Basic Thinking. You can start the day well every morning via our newsletter update.
Openai has released its latest AI model. However, the US group does not want to reveal what the power consumption of GPT-5 looks like. According to experts, it could be significantly higher than with previous models.
With the increasing use of AI systems, global energy consumption increases. According to forecasts, the Electricity consumption of data centers for AI-optimized servers In 2027 there are more than 500 terawatt hours a year. For comparison: In 2023, the number was still around 195 terawatt hours.
Always new progress and further developments in AI systems contribute to this. Because improved skills of AI models also increase the respective resource consumption. Experts estimate that this is also the case with GPT-5 from Openaai. The company does not want to reveal how high the power consumption of the model is.
What is the power consumption of GPT-5?
At the beginning of August 2025 Openai, his AI model GPT-5 released. According to the company Is it “more intelligent in every respect”. The system should provide more useful answers, especially in the areas of mathematics, natural sciences, law and finance. “It is as if you have a team of experts at your disposal that helps you with all questions,” says Openai.
But this further development also has its price. According to experts, a response from GPT-5 can use significantly more energy than an answer to the same question from previous chatt versions.
“A more complex model such as GPT-5 consumes more electricity both during training and during the inference. It is also geared towards long-term thinking,” explains Rakesh Kumar, professor at the University of Illinois, compared to the Guardian. Kumar, who deals with the energy efficiency of computing and AI models, adds: “I can certainly say that it will consume a lot more electricity than GPT-4.”
Energy consumption up to 20 times higher?
That also proves Calculations by researchers from the AI laboratory of the University of Rhode Island. On the day of the publication of GPT-5, these determined the electricity consumption for a medium-long response of around 1,000 tokens. This was up to 40 watt hours of electricity.
The researchers have set up a dashboardin which you document the average power consumption of GPT-5 and compare with other models. Accordingly, the average energy consumption of GPT-5 for a medium-length answer is a little more than 18 watt hours.
GPT-5 is used by “significantly more energy than GPT-4O,” explains Nidhal Jegham, a researcher of the group, compared to the Guardian. Depending on the request, energy consumption could even be up to 20 times higher.
Only recently has Axios foundthat Chatgpt processes 2.5 billion inquiries worldwide every day. Overall, GPT-5 could therefore comparable daily power consumption 1.5 million US households to reach.
Also interesting:
- Artificial hype: GPT-5 is not a new iPhone moment
- Werthms from data centers: Finland shows how energy transition works
- Artificial intelligence: Why AI scares many people
- Study: How artificial intelligence can become more sustainable
The contribution of electricity consumption of GPT-5: As high as from 1.5 million households-first appeared on basic thinking. Follow us too Google News and Flipboard Or subscribe to our update newsletter.
As a tech industry expert, I am not surprised by the high electricity consumption of GPT-5, given its incredibly powerful computing capabilities. The fact that it consumes as much electricity as 1.5 million households per day is a reflection of the massive amount of data processing and computational power required to train and run such a sophisticated AI model.
While this level of electricity consumption may raise concerns about environmental impact and sustainability, it is important to consider the potential benefits and applications of GPT-5. The advancements in natural language processing and AI enabled by models like GPT-5 have the potential to revolutionize industries and improve countless aspects of our daily lives.
As technology continues to advance, it will be crucial for the tech industry to prioritize energy efficiency and sustainability in the development and deployment of AI models like GPT-5. This may involve investing in renewable energy sources, optimizing algorithms for energy efficiency, and exploring new technologies to reduce the environmental footprint of these powerful computing systems. By balancing innovation with environmental responsibility, we can harness the full potential of AI while minimizing its impact on the planet.
Credits