Electricity powered industry.
What could it do for intelligence?
Explore how electricity could become AI knowledge work. Start with real country data, then move the dial to see an illustrative economic scenario.
A thought experiment, not a GDP forecast. Gold bubbles add assumed AI value to observed GDP. This gross scenario does not subtract output lost by diverting electricity from existing uses, or the costs of building and operating AI infrastructure.
Electricity meets prosperity.
↑ Annual electricity generation per person (kWh) · logarithmic scale
Hover, tap, or focus a bubble for details. Select a country above to keep it highlighted. On small screens, swipe the chart sideways.
Shape the scenario Adjust the assumptions behind the gold bubbles
kWh / person × allocation × million tokens / kWh × value / million tokens × realization
The value and realization defaults are editable demonstration assumptions introduced for this explorer. They are not empirical estimates or claims from the infographics. Constant returns are assumed at every allocation level; no adoption timeline, demand saturation, grid losses, displacement, hardware limits, or general-equilibrium effects are modeled. The 100% endpoint is a mathematical limit, not a practical allocation recommendation.
From muscle.
To machines.
To knowledge work.
Coal, steam, and oil expanded the reach of human labor. AI suggests another possibility: electricity translated into reading, reasoning, code, and specialized work. This explorer makes that idea tangible while keeping the assumptions in view.
Each country starts at its observed GDP and electricity generation per person. Allocating a share of that annual electricity to AI creates a hypothetical stream of tokens. An assumed value per token and a realization rate translate that stream into added economic value.
Tokens are not GDP. More generated text does not automatically create useful output. The value assumption is the bridge between technical throughput and the economic scenario; change it to explore how sensitive the result is.
Data, sources & accessible table
All three baseline measures refer to 2023. The embedded snapshot includes 185 countries and territories with positive electricity generation, GDP per capita, and population in that year. Regional aggregates are excluded. Missing observations are omitted, never filled with invented values. Population area is proportional in both bubble layers.
Electricity and GDP: Our World in Data ↗ — Ember (2026) and other electricity sources; Eurostat, OECD, IMF, and World Bank (2026) for GDP. Electricity is generation, not end-use consumption or available surplus. GDP uses purchasing-power parity, in constant 2021 international dollars, not nominal US dollars.
Population: Our World in Data ↗ — UN World Population Prospects (2024) for 2023. Data downloaded October 10, 2026. OWID processing is available under CC BY; underlying sources retain their respective terms. This snapshot does not update automatically.
Inspiration: CirclesX’s “Electricity: The New Barrel of Cognitive Work” and “The Energy Breakout That Multiplied American Wealth” infographics, and a16z’s electricity essay ↗. Historical multipliers and task-hour equivalents are not used as GDP conversion coefficients.
Axes stay fixed while the allocation slider moves, so horizontal movement reflects the scenario. Changing assumptions or scale recalculates the X-axis range. On logarithmic axes, equal distances represent equal ratios.
| Country / territory | Population | kWh / year | Observed GDP | Scenario GDP | Multiplier |
|---|