AI Datacenter Moratorium Simulator Policy Engine

Testing Arvind Narayanan & Ethan Mollick's compute elasticity & algorithmic compounding napkin math

Scenario Presets:

1. Jurisdictional Moratorium & Relocation

% of rejected MW developers rebuild in adjacent friendly states/nations

2. Compounding Efficiency Frontier

Rate inference/compute capability doubles per MW (Chinchilla, pruning, FP4)
Effective Compute Loss
1.25%
0.23 GW Net Lost
Milestone Delay
18.4 Days
~2.6 Weeks
Interstate Spillover
1.28 GW
Exported to other RTOs
Local Grid Relief
1500 MW
Local substation headroom

Interactive Compute Relocation & Capability Trajectory

Policy Divergence: Local Infrastructure vs. Frontier Governance

Policy Dimension Local Moratorium Impact Assessment & Mechanism
Local Grid Strain & Rates High Relief Protects local distribution capacity, avoids immediate substation upgrades and water stress.
Frontier AI Milestone Delay 18.4 Days (<2%) Algorithmic gains & interstate compute reallocation quickly overcome blocked physical capacity.
AI Power Concentration No Effect Hyperscalers simply shift capex to pro-growth states or sovereign international hubs.
Macro AI Governance Negligible (<2% capability shift) Local bans are not an effective substitute for compute governance or alignment treaties.
Enjoy this tool? Build your own with Super