Think of the global economy’s knowledge workers as one giant payroll. Accountants, analysts, lawyers, consultants, coders, marketers: the people who get paid to think for a living. ARK Invest’s Chief Futurist Brett Winton pegs that collective wage bill at roughly $30 trillion (excluding China), and he believes AI is coming for a meaningful chunk of it.
Winton’s thesis is that AI won’t simply make knowledge workers faster. It will begin to substitute for the work itself, redirecting enormous pools of labor spending toward software and automation.
The numbers behind the thesis
ARK projects that AI-driven software spending could land somewhere between $3 trillion and $7 trillion annually by 2030. For context, the entire global software market today is measured in the low single-digit trillions. ARK is essentially arguing that AI alone could double or triple it within half a decade.
One of the most striking data points in Winton’s framework is the cost curve. AI inference costs on agentic benchmarks have reportedly been declining by more than 99% year over year, a pace ARK expects to persist through at least mid-2026.
