Seven core layers. Two frontier theses. One map of where AI's money actually goes. This page explains what that means, what was left out and why, and how to read the judgements behind it.
The ecosystem is described as a city: sand for the chips, power for the energy, land for the data centres, roads for the networking, libraries for the data, brains for the models, shops for the applications, and people for the users who pay for everything. Security (the Guards) and governance (the Rules) cut across all of it rather than forming two more steps.
A metaphor is a teaching device, not a claim about how the industry is actually organised. Real companies operate in several layers at once, supply chains loop back on themselves, and the boundaries between layers are negotiated rather than fixed. NVIDIA designs chips, sells networking and rents cloud capacity. Treat the map as a way in, not as the territory.
Companies appear where they play a genuinely instructive role: a chokepoint, a credible second source, a representative business model, or a useful illustration of how a layer works. This is not an exhaustive directory and it is not a ranking. Many important companies are absent simply because including everything would teach nothing.
Public and private companies are both included and clearly labelled, because several of the most consequential organisations in the ecosystem, OpenAI and Anthropic among them, are private. Where a private company matters, the site notes which listed businesses carry meaningful exposure to it, without implying that buying those businesses is the same thing as owning the private one.
This is a framework for thinking about the economics of a layer, not a recommendation. An important AI company, a strategically advantaged company, an investable security and an attractively valued security are four different things.
Each layer page carries a short framework for thinking about its economics, scored across five dimensions:
Scores are judgements about the layer as a whole, not about any individual company, and they can change quickly. Nothing on this site is investment advice.
Layer pages list the primary sources behind their claims: company investor relations, technical documentation, standards bodies, regulators and research organisations. Specific factual claims carry inline citations back to those sources.
The site was last fact-checked on 29 August 2026. The AI ecosystem changes rapidly. Company positions, technologies and market data reflect information available at the date above. Company positions, tickers, product names and corporate structures are the things most likely to have moved since.
The map is written from the perspective of someone who thinks about capital allocation, which means it overweights structure and underweights the experience of actually building with these tools. Coverage of the ecosystem outside the United States, Europe and Taiwan is thinner than it should be. And any snapshot of a market moving this quickly is out of date somewhere.
Corrections are welcome: info@beaumontmediagroup.com.