All-In Breakdown

AI Energy Consumption

Topic

The significant power demand from training and running large AI models. Hassabis argues that AI's contributions to energy efficiency and new energy sources will outweigh its consumption.

First Mentioned

9/13/2025, 5:47:51 AM

Last Updated

9/13/2025, 5:54:51 AM

Research Retrieved

9/13/2025, 5:54:50 AM

Summary

AI's energy consumption is a significant and growing concern, with projections indicating substantial increases in electricity demand for data centers and AI workloads in the coming years. While current AI models, especially large language models and generative AI, require considerable energy for training and inference, experts like Demis Hassabis of Google DeepMind argue that AI's future contributions to energy efficiency and scientific breakthroughs will ultimately outweigh these costs. The advancement of AI, including technologies like AlphaFold for scientific discovery and drug development, multimodal models like Gemini, and interactive world models like Genie 3 for robotics, is seen as crucial for a 'Golden Age of Science'. However, achieving Artificial General Intelligence (AGI) within the next 5-10 years will necessitate fundamental breakthroughs beyond simply scaling current models, particularly in areas like AI creativity and continual learning, further impacting energy demands.

Referenced in 1 Document
Research Data
Extracted Attributes
  • Environmental Impacts

    Consumption of fossil-fuel-based electricity, significant contribution to greenhouse gas emissions, excessive water consumption for cooling systems

  • Energy Consumption per Text Generation

    Slightly less than 0.5 watt-hours

  • Energy Consumption per AI Server Request

    7-9 watt-hours

  • Comparison to Average Search Engine Query

    ChatGPT-style query is close to ten times more energy than average search; AI server request is 23-30 times more than a normal search (Google 2009 figure)

  • Energy Consumption per ChatGPT-style Query

    Approximately 0.3 watt-hours

  • Projected Annual Growth Rate (AI Energy Consumption)

    Approximately 15% each year from 2024-2030

  • Projected AI Power Demand Increase in UK (next decade)

    500%

  • Projected US Data Center Electricity Consumption by 2030

    7.5% of total US electricity consumption (up from ~2%)

  • Energy Consumption per Image Generation (from text prompt)

    Approximately 0.5 watt-hours

  • Projected Global Electricity Use by Data Centers (2030-2035)

    20% of global electricity use

  • Projected NVIDIA AI Server Consumption by 2027 (Alex de Vries)

    85 to 134 TWh annually

  • Projected US Server Electricity Consumption (High Scenario by 2028)

    Nearly 400 TWh

  • Projected Generative AI Contribution to US Data Center Growth by 2030

    At least 1%

  • Projected Increase in Data Centers and AI Energy Use by 2030 (Rystad Energy)

    Increase by 177 TWh, reaching 307 TWh

Timeline
  • Google reported energy figures for normal search queries, providing a baseline for comparison with AI energy consumption. (Source: web_search_results)

    2009

  • US data centers accounted for 3% of the total US electricity needs. (Source: web_search_results)

    2022

  • AI energy consumption is projected to grow around a steady 15% each year, according to the International Energy Agency. (Source: web_search_results)

    2024-2030

  • NVIDIA servers dedicated to AI could consume 85 to 134 TWh of electricity annually, as estimated by Alex de Vries. (Source: web_search_results)

    2027

  • US server electricity consumption could reach nearly 400 TWh in high scenarios, driven by AI workloads. (Source: web_search_results)

    2028

  • Data centers' electricity consumption is projected to reach 7.5% of total US electricity consumption, with generative AI contributing at least 1% to this growth. (Source: web_search_results)

    2030

  • Data centers and AI energy use are projected to increase by 177 TWh, reaching 307 TWh, according to Rystad Energy. (Source: web_search_results)

    2030

  • US data centers could account for 8% of total electricity needs. (Source: web_search_results)

    2030

  • Data centers could account for 20% of global electricity use, putting immense strain on power grids. (Source: web_search_results)

    2030-2035

  • AI power demand in the UK is estimated to rise by 500%. (Source: web_search_results)

    Within the next decade

  • Artificial General Intelligence (AGI) is predicted to be 5-10 years away, ushering in a new 'Golden Age of Science' and many scientific breakthroughs. (Source: Document 714e6c5f-7b2c-4162-abda-4f48b318c4ed)

    Within the next decade

Web Search Results
  • AI Energy Consumption & Its Impact

    AI energy consumption is projected to grow around a steady 15% each year from 2024-2030 according to the International Energy Agency. Other projections are stating it could go even higher, depending on the rapid development and improvement of data centers and models. Because data centers are running 24/7, they contribute a great deal to grid stress. This can lead to power outages and grid failures, especially in high-demand zones. [...] ### What’s the Shocking Truth Behind AI Energy Consumption? How Much Energy Does AI Use? AI energy usage is at an all-time high, with the average ChatGPT-style query using approximately 0.3 watt-hours. For context, that is close to ten times more than the average search engine query. The estimated cost for the average AI prompt is fractions of a cent. That may not soundRead more “What’s the Shocking Truth Behind AI Energy Consumption?” Read More [...] AI models rely on large amounts of energy to train them. Large-scale data training requires powerful GPUs/TPUs and could take weeks or even months to complete. This results in the consumption of millions of kWh. Once training is complete, the energy use does not stop there; constant responses and searches also expend energy at a staggering rate.

  • Unmasking the Fear of AI's Energy Demand

    Group comes up with similar numbers—electricity consumption by data centers is projected to reach 7.5% of total US electricity consumption by 2030, up from about 2%. Generative AI is expected to contribute at least 1% to this growth. Rystad Energy says that data centers and AI energy use will increase by 177 TWh, reaching 307 TWh by 2030. [...] The bottom line is that we do not need to fear AI’s challenge to the energy grid. Utilities and tech companies will meet increased demand by using a mix of energy sources, including clean and firm electricity supplies like nuclear energy, geothermal power, and even hydropower. AI is not the first—and nor will it be the last—game changer in society’s energy consumption. The discourse on AI's energy footprint must therefore shift from apprehension to proactive problem-solving, focused on energy [...] But beyond the eye-catching statistics, estimates of energy consumption are difficult to find, in part because industry data are heavily guarded and researchers have to rely on overly simplistic extrapolations. The Goldman Sachs Group estimates that AI power demand in the UK will rise 500% over the next decade. U.S. data centers could account for 8% of total electricity needs by 2030, up from 3% in 2022. States that house data centers appear to be running out of power. The Boston Consulting

  • Why AI uses so much energy—and what we can do about it

    Image Image 5: Stacked area and bar chart showing historical and projected U.S. server electricity consumption from 2014 to 2028, broken down by processor type. AI workloads, especially those using 8 GPUs, drive significant growth in projected energy use, with total consumption reaching nearly 400 TWh in high scenarios by 2028. [...] The environmental impact of AI extends beyond high electricity usage. AI models consume enormous amounts of fossil-fuel-based electricity, significantly contributing to greenhouse gas emissions. The need for advanced cooling systems in AI data centers also leads to excessive water consumption, which can have serious environmental consequences in regions experiencing water scarcity. [...] Image Image 4: Quote overlaid on an image of a data center: “By 2030–2035, data centers could account for 20% of global electricity use, putting an immense strain on power grids.” Attributed to Mahmut Kandemir, Distinguished Professor of Computer Science and Engineering. Why is AI’s energy consumption a growing concern?

  • How much energy will AI really consume? The good, the bad and ...

    that each request through an AI server requires 7–9 watt hours (Wh) of energy. That is 23–30 times the energy of a normal search, going by figures Google reported in a 2009 blogpost (see go.nature.com/3d8sd4t). When asked to comment on de Vries’ estimate, Google did not respond. [...] Luccioni and others found that different tasks require varying amounts of energy. On average, according to their latest results, generating an image from a text prompt consumes about 0.5 Wh of energy, while generating text uses a little less. For comparison, a modern smartphone might need 22 Wh for a full charge. But there is wide variation: larger models require more energy (see ‘How much energy does AI use?’). De Vries says that the numbers are lower than those in his paper, but that might be

  • Generative AI: energy consumption soars - Polytechnique Insights

    con­sump­tion by data cen­tres, cryp­tocur­ren­cies and AI could amount to between 160 and 590 TWh com­pared with 2022. This is equiv­a­lent to the elec­tric­i­ty con­sump­tion of Swe­den (low esti­mate) or Ger­many (high estimate). [...] But these fig­ures could sky­rock­et. Alex de Vries esti­mates that by 2027, if pro­duc­tion capac­i­ty match­es the com­pa­nies’ promis­es, NVIDIA servers ded­i­cat­ed to AI could con­sume 85 to 134 TWh of elec­tric­i­ty every year. The cause: the surge in the use of gen­er­a­tive AI. Chat­G­PT, Bing Chat, Dall‑E, etc. These types of arti­fi­cial intel­li­gence, which gen­er­ate text, images or even con­ver­sa­tions, have spread across the sec­tor at record speed. How­ev­er, this type of AI [...] requires a lot of com­put­ing resources and there­fore con­sumes a lot of elec­tric­i­ty. Accord­ing to the AIE, inter­ac­tions with AIs such as Chat­G­PT could con­sume 10 times more elec­tric­i­ty than a stan­dard Google search. If all Google search­es – 9 bil­lion every day – were based on Chat­G­PT, an addi­tion­al 10 TWh of elec­tric­i­ty would be con­sumed every year. Alex De Vries esti­mates the increase at 29.3 TWh per year, as much as Ireland’s elec­tric­i­ty con­sump­tion. “The steady