1 Introduction 4

2 Background: AI’s role in the global economy 5

  2.1 The growing embeddedness of AI into production 5

  2.2 The US as the global leader in frontier AI services 8

3 Model 11

  3.1 Environment 12

  3.2 Domar weights and Hulten’s theorem 12

  3.3 Second-order effects and the post-shock weight 13

4 Empirical Strategy 14

  4.1 Calibrating AI’s pre-shock Domar weight 15

    4.1.1 Global Revenues of the Three Frontier AI Providers 16

    4.1.2 AI’s Domar weight across the basket of countries 18

  4.2 Calibrating the AI Productivity Shock 19

    4.2.1 Estimator A: task-horizon success rates 20

    4.2.2 Estimator B: task mix weighted capability gap 21

  4.3 Calibrating AI’s post-shock Domar weight 22

5 Results: The GDP Impact of AI Access Disruptions 23

6 Conclusion 27