Search Engines vs AI: energy consumption compared

Key statistics

Search engines and AI tools use energy differently, with AI being far more demanding. For example:

• Google Search: Uses 0.0003 kWh per query, emitting 0.2g CO2.
• ChatGPT: Uses 0.0029 kWh per query (10x more energy), emitting 68g CO2.

AI tools like ChatGPT require massive energy for both operations and training, making them less energy-efficient than search engines. As AI adoption grows, balancing progress with energy efficiency becomes crucial.

Comparison of energy use

Metric	Google Search	ChatGPT	Difference
Energy per Query	0.0003 kWh	0.0029 kWh	~10x higher for AI
CO2 Emissions/Query	0.2g	68g	~340x higher for AI
Daily Energy Use	~10.8 MWh	~621.4 MWh	~58x higher for AI
Annual Home Equivalent	~2,000 homes	~21,600 homes	~11x higher for AI

AI’s growing energy needs highlight the importance of designing greener tools to manage digital technologies’ environmental impact.

Search Engine energy use

Search engines depend on servers and data centres to handle tasks like web crawling, indexing, and processing search queries. These operations require a lot of computing power and cooling systems, making them energy-heavy.

For example, Google’s data centres process about 40,000 search queries every second, showcasing the immense scale of their energy demands.

Search Engine carbon footprint

According to FusionChat, each Google search uses 0.0003 kilowatt-hours (kWh) of energy, resulting in 0.2 grams of CO2e emissions. While the impact of a single search seems minor, it adds up quickly, given the billions of Google searches conducted every day.

Activity Type	Energy Consumption	CO2e Emissions
Single Google Search	0.0003 kWh	0.2g CO2e
Daily Google Searches (Global)	~10,800 kWh	~6.9 metric tons CO2e

Companies are improving server and cooling systems, adopting renewable energy sources, and refining search algorithms to lower energy use and emissions.

While search engines have made progress in efficiency, the growing presence of AI tools within search results adds new complexity to the conversation about digital energy consumption in search results.

AI energy use

AI energy use for operations

AI tools like ChatGPT rely on substantial computational power to function. For example, each query uses about 0.0029 kWh of energy. With approximately 200 million queries processed daily, this translates to an operational energy consumption of 621.4 MWh – comparable to the yearly electricity usage of 52 average American households.

Operation Type	Energy per Query	Daily Energy Use
ChatGPT Response	0.0029 kWh	621.4 MWh
Google Search	0.0003 kWh	~10.8 MWh
Energy Difference	~0.0026 kWh	~610.6 MWh

Luccioni, S., Jernite, Y. and Strubell, E., 2024, June. Power hungry processing: Watts driving the cost of AI deployment?. In The 2024 ACM Conference on Fairness, Accountability, and Transparency (pp. 85-99).

AI energy use for models and training

The energy required to train AI models is even higher than for daily operations. For instance, training GPT-3 consumed 1,287 MWh, while GPT-4 training demanded 1,750 MWh. These figures underscore how energy-intensive it is to develop advanced AI systems.

The energy consumption of both operating and training AI tools highlights the importance of finding ways to make these systems more efficient. Compared to search engines with a much lower energy footprint, the growing demand for AI calls for a careful balance between advancing technology and addressing environmental concerns.

Search Engines vs AI: energy comparison

Energy use and emissions comparison table

Metric	Google Search	ChatGPT	Impact Factor
Energy per Query	0.0003 kWh	0.0029 kWh	About 10x higher for AI
Daily Queries	Billions	200 million	–
Daily Energy Use	~10.8 MWh	621.4 MWh	About 58x higher for AI
Annual Home Equivalent	~2,000 US homes	21,602 US homes	About 11x higher for AI
Scalability Efficiency	High	Limited by energy demands	–

This table underscores the energy gap between traditional search engines and AI tools, highlighting the growing need for solutions that address their environmental demands.

AI energy use and sustainability impact

Projections suggest that AI energy consumption could surpass France’s total energy usage by 2030.

Stanford University calculated that fine-tuning GPT3, took 1,287 MWh (Megawatt hours) of electricity. In the United States, where many servers used for AI are located, 16% of electricity still comes from coal—the dirtiest source of energy. Training GPT3 in that way caused 502 tonnes of CO2, which is as many emissions as driving a car to the moon and back.

AI technology like GPT requires exponentially more energy with each new development step.

AI tools face a scalability hurdle because their energy needs increase significantly with usage. In contrast, search engines have refined their processes to operate more efficiently. To address these challenges, sustainability efforts are currently focused on:

• Improving infrastructure to reduce energy consumption
• Incorporating renewable energy sources into operations
• Designing AI systems that require less energy to function effectively

Tools for AI and machine learning platforms

CodeCarbon zeroes in on emissions from AI and machine learning processes, addressing the high energy demands of these technologies.

AI & Sustainability Lead at Hugging Face, Dr Sasha Luccioni highlights the importance of careful AI use:

This perspective reinforces the role of emission-tracking tools in managing computational resources thoughtfully.

FAQs about AI and Search energy usage

How much CO2 is emitted by a Google search?

Digital tools, while convenient, come with their own environmental costs. For instance, running a single Google search emits about 0.2g of CO2. To put things into perspective, here’s a comparison of the energy use and emissions between Google searches and AI queries:

Activity	Energy Consumption	CO2 Emissions
Google Search	0.0003 kWh	0.2g CO2
ChatGPT Query	0.0029 kWh	68g CO2

As noted earlier, AI systems like ChatGPT consume significantly more energy than traditional search engines. This increased energy demand has contributed to rising emissions. For example, according to NPR, Google’s emissions have surged by 48% since 2019, and Microsoft’s emissions have grown by 29% since 2020.

To address these challenges, organizations can use tools such as CodeCarbon to track and reduce the carbon footprint of AI systems. If left unchecked, AI’s energy consumption could double France’s current energy usage by 2030, emphasizing the urgency for sustainable approaches.

Back to News & Insights