Answers are grounded in this paper and cite their sources.
This report examines where the economic profit created by artificial intelligence is likely to settle. Its central claim is that a great technology is not automatically a great investment: the inventor, the infrastructure builder, the model provider and the eventual profit winner need not be the same company. The analysis follows a chain from power and data centres through chips, networking, cloud, models and software to the businesses and consumers who must ultimately pay, asking at each layer who pays whom and who keeps the incremental return after competition and financing costs. A central concern is the bridge between productivity and purchasing power: productivity creates capacity, but it does not by itself create customers. The report therefore maintains a Demand Ledger tracking whether the gains from AI reach the households and institutions that ultimately buy things, and a Scarcity Ledger tracking where rents may migrate once intelligence becomes abundant. Company evidence is drawn from a six-firm sample of first-wave AI leaders using issuer annual reports, SEC filings and investor-relations releases only, so that every figure can be reproduced without a proprietary data licence. Operating cash flow and property-and-equipment spending are shown separately rather than relying on any single free-cash-flow definition. The evidence shows an asymmetry: suppliers convert a large share of revenue into cash while spending comparatively little on their own fixed assets, whereas the hyperscalers are absorbing tens of billions of dollars a year of infrastructure cost whose lifetime return is not yet demonstrated. Conclusions are graded Supported, Developing, Mixed or Insufficient Evidence, and can move between those states as data arrive. The report states explicitly what remains unknown and specifies in advance the evidence that would prove its central concerns wrong.
No takes yet. Share an insight, caveat, or question.
Larry Lim Kheng Cheong (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: