A new Bank of Canada staff working paper, “Measuring the AI Economy,” raises an important measurement issue: AI may be growing much faster than traditional economic statistics can easily capture.

The authors estimate that U.S. AI compute spending grew by more than 140% per year in both 2024 and 2025. Physical compute capacity grew by more than 200% per year, while quality-adjusted AI production grew by more than 2,000% per year.

In practical terms, “quality-adjusted AI production” means the authors are not just counting dollars spent on AI systems. They are trying to estimate how much more AI capability is being produced as chips improve, models become more efficient, and the cost of producing similar AI outputs falls.

Their proposed “AI GDP” framework is still early and comes with significant caveats, but it points to a real challenge for economic measurement: AI-related activity is spread across data centres, cloud computing, software, professional services, R&D, and end-user applications.

That makes AI difficult to see as a coherent sector.

For local and regional economies, including Huron County, the implication is less about the headline number and more about preparedness. If AI capabilities continue to improve rapidly, businesses may feel the effects through changing productivity expectations, new service models, shifting workforce needs, and different forms of competition before those changes are obvious in standard economic reports.

The paper is also careful to distinguish between AI’s productive capacity and its realized value for people and businesses. For employers, communities, and policymakers, the practical question is not just how powerful the technology becomes, but how broadly and usefully it is adopted.

Read the Bank of Canada staff working paper: “Measuring the AI Economy.”
https://www.bankofcanada.ca/2026/06/staff-working-paper-2026-20/