The Limits of Efficiency: Organizational Failure, Global Crises, and the Physical Constraints of AI Capital / 效率的边界:从组织失灵、全球危机到人工智能资本的物理约束
This working paper asks what happens when efficiency is pursued as an organization's supreme criterion over a long period, and treats the question as empirically testable rather than normative. The design is a three-tier test sharing one attribution framework of seventeen failure types and a three-level evidence hierarchy (adjudicated / verified / estimated): (1) Organizational tier — 29 organizations that practised efficiencyism for a decade or more (1913–2026). Of the six that perished, five involved fraud or leverage; a third of the sample sits in intermediate states rather than at either extreme. (2) Systemic tier — 26 global financial and technological crises (1997–2026), scored on eight structural dimensions. Governance failure and complexity/opacity each appear in 23 of 26 crises, while leverage ranks last (15 of 26): leverage is an amplifier, not a cause. (3) Contemporary tier — a physical audit of AI capital commitments (2026). Roughly USD 1.2 trillion of binding contracts are constrained not by capital but by a 61-month interconnection queue, gas-turbine slots sold through 2030, and 160-week transformer lead times. The most prominent commitment, Stargate, has realised about 6% of announced capacity. The synthesis proposes that technological diffusion and corporate failure are two sides of one process; that the financial and physical time constants differ by an order of magnitude; and that if AI becomes infrastructure, the compute layer is the most likely candidate for utility-ization or sovereignisation. Contains 4 figures, 3 tables, 25 references, and six explicitly falsifiable propositions. 本文件含中文版与英文版各一份:《效率的边界:从组织失灵、全球危机到人工智能资本的物理约束》/ The Limits of Efficiency。
Authors
- jikko Fu
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-20
- DOI
- https://doi.org/10.5281/zenodo.24397570
- Primary Topic
- Impact of AI and Big Data on Business and Society
- Type
- article
- Field-Weighted Citation Impact
- 0.00