AI economy grew over 2000% in 2024-25, new GDP framework proposed

A Bank of Canada working paper reveals the rapid expansion of the AI economy, estimating that nominal AI compute spending grew over 140% per year in 2024 and 2025.

Raw compute capacity saw over 200% annual growth, while quality-adjusted AI output surged by more than 2000% per year.

These rates reflect three compounding forces: expanding data-center capacity, continued improvements in chip efficiency, and rapid algorithmic progress.

The authors note that traditional economic statistics struggle to capture this activity, as AI's impact is scattered across various industry categories, quality improvements are unusually rapid, and its role in the economy is evolving beyond a passive capital input.

This makes tracking the AI economy as a coherent whole challenging within existing national accounts.

The paper introduces a nascent framework for "AI GDP" to provide an integrated picture of the AI economy's size and growth.

This approach tracks all compute production, model development, and inference output, independent of traditional industry classifications.

It applies granular quality adjustments, with the inference deflator declining approximately 94% per year, reflecting a 16-fold increase in quality-adjusted output for a given dollar spent.

The framework also treats AI training as investment in "model capital," aligning with national accounts' capitalization of R&D.

This complementary perspective aims to provide early signals of AI's growing economic role.