B2B does not create an independent economy
The household constraint can be overstated if it is expressed as 'all B2B spending comes from households.' That is literally false. Firms invest, governments spend, capital formation creates future capacity, and intermediate transactions can be enormous. The stronger statement is about economic closure.
Intermediate expenditure must ultimately justify itself through useful final output or sustainable future productive capacity
A chip manufacturer sells to a cloud provider, which sells to a software company, which sells to a manufacturer, which sells to another business. None of those transactions requires immediate household consumption. Yet an indefinitely expanding chain of intermediate spending cannot create value merely by circulating invoices. The capital must eventually increase the capacity to produce something someone values, reduce the resources required to produce it, or create an asset expected to do so later.
This distinction is important for AI infrastructure. A large data-center buildout can be rational even before final demand is fully visible if it creates productive capacity with valuable future uses. But capex itself is not proof of the revenue required to justify the capex. The analysis must close through future cash-generating output.
Capex today → productive capacity → future useful output → monetizable demand → cash flow
AI can sit far upstream from households and still be constrained by this chain. Moving the provider upstream does not free it from economic closure; it changes the timing and form of the claim.