Skip to main content
The Efficiency Leaderboard

Low-power lifestyle.
No succulents were harmed in the making of this.

A live ranking of how much water, energy, and carbon are used for everyday life — and why the hardy succulent is the undefeated champion.

Since you opened this page
AI queries have used roughly 0 litres of water.
Your succulent used 0.
Time machine
2025
today
2015202020252030

Today's baseline. Now slide either way.

🏆 The Champion
🌵

Hardy Succulent

A+

Misted monthly. Sits in natural light. Judges you silently.

Thirst
<0.01 L
Energy
0 kWh
Carbon footprint: 0 g CO₂
📺
#2

Video Streaming

We think YouTube and others use a lot more energy and water than this — but we don't have data to show this.
D

A binge-watch is a small data-centre marathon.

Thirst
0.91 L
Energy
0.11 kWh
Carbon footprint: 48.0 g CO₂
🤖
#3

Reasoning LLM Query

F

Each thought cycle is cooled by data centre water. You can almost hear it sweat.

Thirst
15.0 L
Energy
0.09 kWh
Carbon footprint: 129.0 g CO₂
#4

Golf Course (Av hourly usage)

G

A green desert. Drinks more in a day than you will all year.

Thirst
46.6k L
Energy
33.9 kWh
Carbon footprint: 10.6k g CO₂
✈️
#5

Aeroplane Flight

G

One hour airborne ≈ a year of your succulent's entire footprint × infinity.

Thirst
52.5 L
Energy
294.0 kWh
Carbon footprint: 94.5k g CO₂

Head to Head

Contender
vs
Compared to
Water
7,200×
AI Query drinks 15.0 L vs Succulent's <0.01 L
Energy
AI Query burns 0.09 kWh vs Succulent's 0 kWh
Measuring this page's footprint…

Every figure on this page traces back to a real source. We mark live data clearly, flag estimates honestly, and tell you what we're upgrading next.

AI query water use (~500ml per reasoning query)
Estimated
Li et al., "Making AI Less Thirsty" (UC Riverside / U. Texas)
Peer-reviewed estimate. Live AI water-use APIs do not yet exist publicly.
AI query energy use (~3 Wh, reasoning models higher)
Estimated
University of Michigan ML Energy initiative
Streaming hour (~0.1 kWh, ~45 g CO₂, ~0.85 L water — combined data-centre + network + device)
Estimated
Midpoint of published estimate ranges (IEA, Carbon Trust, Netflix sustainability)
Single point estimate. Real per-hour impact varies by codec, device class and grid mix; YouTube and other ad-supported platforms are likely higher.
Golf course daily water use (~290k US gallons / 1.1M litres)
Estimated
GCSAA — Golf Course Environmental Profile
Average across US courses; arid-region courses use significantly more.
Short-haul flight CO₂ (~90kg per pax-hour)
Planned
Carbon Interface API (template ready)
Wire-up template is in the code (see useSiteFootprint section). Drop your API key into env vars and uncomment.
This page's own footprint (live)
Live
PageSpeed Insights API + browser fallback
Tries the /api/page-footprint route first (Lighthouse-grade, cached 24h). Falls back to browser Performance API. Energy/water conversions via Sustainable Web Design model.
Hardy succulent: 0L water today, 0 kWh
Live
Benchmark — verified by static physical reality
A succulent in our shop window does, in fact, use zero kWh. We are confident in this reading.
AI growth trajectory (year-curve)
Estimated
IEA Electricity 2024 + IEA Data Centre forecasts
Illustrative projection based on data centre demand forecasts.

Pick the winner.

Bring home a hardy succulent. It'll outlive your last three phones.

Shop succulents
Figures are research-backed estimates (IEA, GCSAA, Carbon Interface, U. Michigan). Year-curve trajectories are illustrative — the AI growth curve in particular is a defensible rough projection. Data layer is structured for swap-in to Climatiq, Carbon Interface, and shop smart-meter APIs.
Hardy Succulents net-zero plan