The First Concentration
The attention economy was built by learning how to predict what a person would watch, click, share, buy, and return to. Recommender systems turned entertainment, news, shopping, and social life into optimized feeds.
At first this looked like convenience. The song found you. The video played next. The product appeared. But over time, the user stopped being the editor. The system selected more of the menu.
Leisure screen hours: chosen vs. served
Daily leisure screen hours split between recommender-driven and non-recommender use.
The Cash Machine
Attention produced enormous cash flows. Consumer platforms reached scale, monetized the attention layer, and generated operating margins that few industries could match. That cash went partly to shareholders, partly to infrastructure, and now increasingly to AI.
This is the loop: attention produces revenue; revenue funds compute; compute improves prediction and generation; better prediction and generation capture more attention and more markets.
Hyperscaler capex, 2020-2026
Microsoft, Alphabet, Amazon, and Meta aggregate capital expenditure.
The Second Concentration
The intelligence economy is not separate from the attention economy. It is downstream of it. The firms with data centers, distribution, identity systems, app ecosystems, cloud platforms, ad networks, and consumer relationships are positioned to turn AI into everyday infrastructure.
That matters because the same incentive problems can move upward. If intelligence tools are funded by attention capture, they may be optimized for dependency, retention, and control rather than independence, skill, and choice.
Break The Loop
American Fabric's constructive challenge is to break that loop. Intelligence tools should help people think, work, learn, create, and choose with more agency. They should not simply become the next layer of behavioral capture.
The future of AI will be shaped not only by model capability, but by who owns the interface between intelligence and daily life.