Your Brain Runs on 10 Watts — Less Power Than Most Light Bulbs

A 1.4-Kilogram Organ That Runs the World

The human brain weighs approximately 1.4 kilograms — about 2 percent of average body weight — yet consumes roughly 20 percent of the body's total energy budget at rest. This disproportionate energy demand has been the subject of intensive research, but the absolute power consumption remains modest by any engineering standard: approximately 10 to 20 watts, with the most common figure cited in neuroscience literature being about 12 watts for a resting, alert brain.

This 10-to-20-watt figure is not evenly distributed. Different brain regions consume different amounts of energy depending on their activity level. The frontal cortex, responsible for executive function and planning, uses more energy during complex cognitive tasks. The visual cortex spikes in activity when processing complex scenes. Baseline activity — the default mode network active during daydreaming and rest — maintains constant background processing that accounts for a large fraction of the brain's total consumption even when no particular task is being performed.

What 10 Watts Actually Buys in Computing Terms

The comparison between brain power consumption and artificial computing power is where the 10-watt figure becomes genuinely staggering. The brain contains approximately 86 billion neurons, each making between 1,000 and 10,000 synaptic connections with other neurons. The number of synaptic connections in a human brain is estimated at between 100 trillion and 500 trillion. Each synapse processes signals and modifies its response based on history — a form of computation that is distributed across every connection in the network simultaneously.

Current computer models that attempt to simulate brain-scale neural networks require hardware consuming hundreds of kilowatts to petawatts of power for partial simulations of brain regions. The Human Brain Project's simulations of portions of the cortex required supercomputer clusters consuming millions of watts to model minutes of brain activity at reduced fidelity. The brain performs the real thing, continuously, in real time, on 10 watts.

The efficiency advantage comes from the physical nature of synaptic computation. Biological neurons communicate through electrochemical signals that require very small amounts of energy per transmission — on the order of 10^-14 joules per synaptic event. Silicon transistors switching at comparable rates require orders of magnitude more energy per operation. The brain's computational substrate is simply much more energy-efficient than any silicon equivalent yet built.

Why the Brain Is So Efficient

Several factors contribute to the brain's extraordinary energy efficiency. Neurons operate asynchronously — they fire only when needed rather than cycling continuously like transistors in a clock-driven processor. The analog nature of neural computation allows graded responses rather than binary all-or-nothing operations, packing more information per energy expenditure. And the massively parallel architecture means that complex computations are distributed across billions of simultaneous processing units rather than executed sequentially by a small number of fast processors.

The brain's energy efficiency is also shaped by evolution's constraint: for most of human evolutionary history, food was often scarce, and any organ consuming more energy than necessary was a metabolic liability. The brain's computational power represents a genuine energy bargain — maximum capability per calorie.

The Implications for Artificial Intelligence

The brain's 10-watt benchmark has become one of the primary targets of neuromorphic computing — the engineering field dedicated to building computer hardware that mimics the architecture of biological neural networks rather than conventional processor designs. Intel's Loihi chip and IBM's TrueNorth project are examples of neuromorphic hardware designed to approach brain-like energy efficiency. None has yet come close to matching the full capability of the brain at anything approaching 10 watts, but the target is clear and the engineering motivation compelling: if a computer could do what the brain does on the same power budget, it would represent one of the most consequential technology advances in human history.