Independent research on AI economics and infrastructure.
AI is attracting enormous investment while changing the economics of the businesses expected to pay for it. Threadbaire investigates where the value is actually being created, who gets to keep it, and who ends up paying for it.
By Lida Liberopoulou · All research CC BY-SA 4.0
The companies building AI are increasingly also financing, supplying, and buying from one another. These relationships make it difficult to distinguish independent demand from activity financed within the same ecosystem, or to see where financial risks ultimately accumulate.
Threadbaire examines how these dependencies formed, what sustains them, and where the costs and risks might land when those relationships come under pressure. The research draws on earlier infrastructure industries and the evolution of the software economy to identify patterns that company-by-company analysis can miss.
The foundational argument
Why AI is undermining the economics of the software industry, and why attempts to preserve existing business models may deepen the problem.
The structural and historical framework
How AI's industrial costs collide with eroding revenue, how computing once escaped the historical infrastructure pattern, and how the conditions of that escape were gradually reversed.
Part 1: AI's Gilded Age · Part 2: Tech's Great Escape · Part 3: The Return of the Bind
The applied analysis
AI financing relationships read through corporate disclosures, with claims recorded before the earnings and then checked against the filings.
Five predictions and one control, filed before Microsoft, Alphabet and Amazon reported, scored against their filings. Four held, one could not be tested. The more useful answer sits in the notes, where more of the bill now lives: guarantees, collateral, and companies these firms part-own.
31 October 2026
Alphabet and Meta have each spent more cash in a quarter than their businesses brought in. Before Microsoft, Alphabet and Amazon report on 28 and 29 October, five checkable claims on one question: if their cash looks better, is it real, or has the bill moved somewhere else?
27 October 2026
Anthropic is about to make its S-1 public. Most of its big numbers have leaked. Four checkable predictions about whether it will let a reader follow the money, filed before it appears.
23 October 2026
I take a small number of commissions. Each is an independent, written analysis of a specific AI exposure, thesis or decision: what the evidence shows, what it depends on, and what to watch.
Threadbaire combines economic and historical analysis with direct experimentation using the systems it examines. Working across AI models reveals practical dependencies and limitations that aren't always visible from outside. The research uses an open methodology that keeps context portable and decisions traceable across tools.
The method and templates are freely available.
