AI Search vs Traditional Search: Electricity, Water & Carbon Index 2026
The environmental cost of AI-generated answers versus conventional search, benchmarked for UK decision-makers. An independent, evidence-based comparison of electricity, water and carbon per query across Google, Gemini, ChatGPT, Copilot, Perplexity and Claude — with forecasts to 2035.
What this report covers
A single AI-generated search response consumes an estimated 3 to 100 times more electricity than a traditional Google query, depending on model, query complexity and data centre location. This report benchmarks the electricity, water and carbon footprint of leading AI search platforms against conventional search, quantifies grid and water impacts, and projects the trajectory to 2035 under three scenarios.
Headline conclusions
- A traditional Google query uses ~0.3 Wh; disclosed AI queries range from 0.24 Wh (Gemini) to ~2.9 Wh (long-form ChatGPT).
- Global data centre electricity reached ~415 TWh in 2024 and is projected to nearly double to 945 TWh by 2030, with AI the primary driver.
- Data centre electricity demand grew 17% in 2025 — nearly six times the 3% growth in overall global electricity demand.
- Global data centre water use is ~560 billion litres per year and could exceed 1 trillion litres by 2030.
- Only Google and OpenAI have published verified per-query energy figures — the wider AI industry remains opaque.
Intended audience
- Policymakers, regulators and DESNZ / Ofgem analysts
- Sustainability, ESG and net-zero leads at UK enterprises
- Search, marketing and digital leaders assessing AI adoption
- Researchers, journalists and civil-society organisations
- Data centre operators, hyperscalers and grid planners
Inside the 44-page report
- 011. Executive summary — the environmental cost of the search transition
- 022. Methodology — measuring electricity, water and carbon per query
- 033. Data centre operations — infrastructure behind every query
- 044. The evolution of search — from directories to generative AI
- 055. How traditional search works — crawling, indexing and ranking
- 066. How AI search works — LLMs, GPUs and inference
- 077. Electricity consumption — platform-by-platform comparison
- 088. The 10× to 100× energy gap in context
- 099. Water consumption — the hidden thirst of AI search
- 1010. Cooling technologies and regional water-stress implications
- 1111. Carbon emissions — grid mix, renewables and Scope 2 reality
- 1212. Data centre growth and grid pressure
- 1313. Economic and productivity impact of AI search
- 1414. The future of search — agentic AI energy concerns
- 1515. International comparison — UK, EU, US, China, Japan, Singapore
- 1616. UK deep dive — demand, grid queue and policy response
- 1717. Forecasts to 2035 — electricity, water, carbon and query volume
- 1818. Recommendations for government, industry and users
- 1919. Balanced assessment, glossary, references and appendices
Sample pages
Frequently asked questions
Is the report free to download?+
Yes — the full 44-page report is free. We ask for an email so we can send you future updates and related research.
How do you compare energy per query fairly?+
Disclosed corporate figures (Google, OpenAI) are reported separately from third-party academic estimates for Copilot, Perplexity, Claude and Meta AI. Assumptions, ranges and uncertainty are documented in full.
Does it cover water and carbon as well as electricity?+
Yes — the report benchmarks Water Usage Effectiveness, Scope 2 emissions and regional grid intensity alongside per-query electricity.
Download the full report
Enter your email to get the full PDF plus the weekly UK AI energy briefing.