When you type a question into an AI chatbot, the response arrives in seconds. It feels effortless, like the machine just knows things. But behind that response is a supply chain that touches three categories of shared resources, none of which appear on your bill.
The electricity powering the data center comes from the Earth. The knowledge the model was trained on was accumulated over centuries by millions of people. And the query you just typed is itself a data point that improves the model and increases the company’s valuation.
Earth. Light. Signals. Every product in the modern economy draws on all three. This post explains what they are, why they matter, and why nobody is charging for them properly.
Earth: The Finite Ground We Draw From
Earth is the physical commons. It includes minerals, metals, water, timber, fisheries, the atmosphere, orbital slots, and the electromagnetic spectrum. These are inputs that exist before any company touches them.
AI makes the scale visible. Global data center electricity consumption is projected to reach 1,050 terawatt-hours by 2026, roughly what Japan uses in a year. AI operations alone are expected to consume 90 TWh annually by that date, a tenfold increase from 2022. Google plans to spend $75 billion on AI infrastructure in 2025. That infrastructure runs on copper, silicon, lithium, cobalt, water for cooling, and land for the buildings.
The DRC supplies over 70% of the world’s cobalt, a critical input for batteries that power everything from electric vehicles to backup systems in server farms. Thousands of people, including children, mine it under dangerous conditions. The cobalt leaves the DRC. The batteries go into devices and data centers around the world. The revenue stays with the companies that buy it on the open market. The Congolese miners do not receive a share of the value their resource generates downstream.
The atmosphere absorbs the emissions from all of this. Carbon pricing instruments generated over $100 billion in 2024 from 80 jurisdictions. Spectrum auctions have raised over $233 billion. The Earth commons is already being priced. The revenue just does not reach the owners.
Light: The Knowledge We Inherit
Light is the intellectual commons. It includes scientific research, mathematical principles, open-source software, engineering standards, medical knowledge, and the accumulated written record of human civilization. These are inputs no one alive invented from scratch. They were built across generations.
AI makes this dependency obvious. Large language models are trained on enormous datasets scraped from the internet: books, research papers, Wikipedia articles, news archives, forum posts, open-source code. OpenAI was ordered by a federal judge to reveal internal communications about two datasets of pirated books downloaded from Library Genesis. Multiple companies, including OpenAI, xAI, and Google, face lawsuits from authors alleging unauthorized copying of copyrighted works to train their models.
The legal battles are about copyright. The economic question is broader. Even the material that is not copyrighted, the public-domain research, the freely shared code, the mathematical foundations, represents centuries of collective intellectual work. No company built calculus. No startup invented the scientific method. These are inherited inputs, and they are the reason AI works at all.
When a company trains a model on this accumulated knowledge and charges for access to it, the Light commons is being monetized. The question is whether any of that revenue should flow back to the commons that made it possible.
Signals: The Data We Generate
Signals is the behavioral commons. It includes the clicks, searches, purchases, movements, sensor traces, and interactions that billions of people generate every day. This data is the fuel that powers targeted advertising, recommendation algorithms, and AI training.
The global advertising industry generated $1.14 trillion in 2025. That revenue depends almost entirely on behavioral data collected from users. Google and Meta do not sell ads into a vacuum. They sell access to detailed profiles built from what people do online. The data is the product. The people who generate it are not compensated.
AI adds another layer. Every query you type into a chatbot, every correction you make, every thumbs-up or thumbs-down you click, feeds back into the model. Your interaction makes the product better. The company captures that value as improved performance, which drives subscriptions and enterprise contracts. You contributed a signal. You received nothing for it.
This is not a hypothetical problem. The EU’s General Data Protection Regulation gives individuals rights over their personal data. California’s Consumer Privacy Act does the same. But data protection is not the same as data compensation. The right to delete your data is not the same as the right to a share of the revenue it generates.
Why All Three Matter Together
Most policy discussions treat these inputs separately. Environmental policy handles the atmosphere. Intellectual property law handles knowledge. Privacy regulation handles data. But the economy does not separate them. Every product draws on all three at once.
An AI query uses electricity from the Earth, knowledge from the Light, and your behavioral data as a Signal. An electric vehicle uses lithium and cobalt (Earth), battery chemistry developed through decades of public research (Light), and driving telemetry that improves autonomous systems (Signals). A streaming service runs on spectrum (Earth), a catalog of human creative work (Light), and viewing data that shapes what gets produced next (Signals).
The three commons framework treats them as a unified set of shared inputs that generate private returns. The question for each one is the same: is the rent being collected, and if so, where does it go?
Right now, some of it is collected. Carbon is priced in 80 jurisdictions. Spectrum is auctioned in over 100 countries. But almost none of the revenue reaches the people who share these inputs. And the Light and Signals commons are barely priced at all.
The framework proposed in Shareholder at Birth argues that all three commons should be priced at existing gates, the revenue pooled in a public trust, and the returns distributed equally to every person alive. The logic is the same one behind spectrum auctions and carbon pricing: if a company profits from a shared input, it pays rent. The missing step is making sure that rent reaches the people who own the input.
