Every infrastructure technology before AI followed the same pattern. The question is if AI will continue or break it.
Threadbaire is an independent research practice publishing structural analysis of AI infrastructure economics. The work is grounded in direct experience building with these systems (both open and proprietary). Current focus: the economic forces that push AI toward consolidation, the institutional conditions that historically prevented it, and what their absence means now.
By Lida Liberopoulou· All research CC BY-SA 4.0
Available for advisory, research collaboration, and expert consultation.
Six structural claims about hyperscaler earnings disclosure, filed before the event with pass/fail conditions, all confirmed. The score matters less than what it reveals: a recurring pattern in which investment effects, commercial demand, and infrastructure costs are each visible, but the connections between them are not.
31 July 2026
Nvidia is guaranteeing its own customer's debt. Google is booking unrealised gains on companies that buy its cloud compute. Amazon's Anthropic position contributed $12.3 billion to quarterly income from a price adjustment. On Wednesday and Thursday, Microsoft and Apple report from opposite sides of this pattern. Six falsifiable predictions about what the filings will disclose and what they will leave impossible to calculate.
27 July 2026
Microsoft voluntarily disclosed Xbox's internal margin after twenty-five years of absorbing worse losses quietly. The question is not whether Xbox deserved a reset but why the only window into internal economics opened for the business unit furthest from the question investors are actually asking.
3 July 2026
The analytical framework centres on the tech infrastructure double bind. It is the combination of the industrial infrastructure costs that cannot be avoided with value dissolution caused by the technology that erodes the revenue base funding the buildout. The work traces this mechanism through historical infrastructure industries (railroads, electricity, telephony) and applies it to AI's current trajectory, mapping how circular financing, platform capture, and institutional voids produce the concentration patterns visible today.
The foundational argument is in the thesis. The blog extends it through the Silicon Industrialists series (a tetralogy on the economic forces shaping AI), the agent web triptych (credential, commercial, and memory capture), and standalone analyses of model commoditisation, labour market effects, and software category economics.
The research is produced using an open methodology for AI workflows (portable context, model-agnostic setups, and receipts-based decision tracing across tools). The method is documented publicly and the templates are free. If you work with AI across multiple models and want your reasoning traceable and your context portable, the method page explains how it works and the templates are on GitHub.
