EconomicsFeatured paper
Allocating Sovereign AI Compute: A Mechanism-Design Audit, an Impossibility, and a Credit-Market Response
Abstract
This paper audits how major sovereign AI-compute programmes allocate scarce accelerator capacity and finds that none elicits cardinal preferences through a formal mechanism. It then develops a two-stage credit mechanism in which a planner assigns endowments from a public-value assessment before a VCG or clinching auction is run in credits. A worked divisible instance compares six allocation mechanisms and shows how the policy recommendation changes when public programmes prioritize distributional objectives rather than willingness-to-pay efficiency.
Publication Metadata
Keywords
mechanism designbudget constraintscompute governanceVCGfair divisionAI infrastructure
Citation
Christian Hansen (2026). Allocating Sovereign AI Compute: A Mechanism-Design Audit, an Impossibility, and a Credit-Market Response. The Insight Research. Sovereign AI Compute manuscript, August 2026
Related Papers