The sell-side consensus on AI infrastructure prices capex in absolute terms — six hundred billion here,seven hundred billion there — and prices inference output in narrative terms — “the price of intelligence iscollapsing.” Both framings miss the same thing. Inference output is a manufactured good. Its marginal costis dominated by electricity. And electricity is not priced globally, it is priced by interconnect.This paper develops a four-tier framework for the geographic pricing of AI inference between 2026 and2030. Tier 1 (Sovereign Capable) — China, the Gulf, the Nordic countries, France — combines lowmarginal electricity cost with state-backed capacity expansion. Tier 2 (Emerging Stretched) — India,South Korea, Japan — has committed hundreds of billions to sovereign compute on grids that are not yetstable enough to clear it. Tier 3 (Power Constrained Wealthy) — the United States in the PJMinterconnect, Germany, the United Kingdom — concentrates capex without the physical electricity todispatch it; PJM has now cleared its annual capacity auction at or near the regulatory cap three years in arow, and the 2027/2028 auction came in short of its reliability target for the first time in the RTO’s history.Tier 4 (Arbitrage Suppliers) — Russian Siberia, Canadian hydro, the Brazilian Norte/Nordeste, AustralianPilbara — owns the cheapest electrons on earth and is now exporting them, increasingly, to Tier 1workloads via cross-border power agreements and dual-resident cloud structures.
Research et al. (Thu,) studied this question.