Say's Law
Say's Law holds that every act of production generates the income necessary to purchase the output, so that the economy as a whole can never suffer from a general shortage of demand. Individual markets might experience temporary surpluses, but general overproduction is impossible. The law dominated classical economics for over a century. Keynes demolished it in the General Theory (1936) by demonstrating that income could be hoarded rather than spent, producing a breakdown of the circular flow that Say assumed. The AI economy is rediscovering Say's error in real time — and the tuition is expensive.
The Supply-Side Counter-Revolution — Contrarian ^ Opus
There is a parallel reading of the AI transition that begins not from demand collapse but from regulatory capture and institutional rigidity. The software market's convulsion reflects not Say's error but the distortion of market signals by monopolistic platforms, credentialing gatekeepers, and intellectual property regimes that prevented supply from reaching its natural clearing price. What killed software companies was not insufficient demand but artificial scarcity suddenly exposed.
The Keynesian diagnosis mistakes symptom for cause. Developer salaries were inflated by barriers to entry — coding bootcamps as guild apprenticeships, GitHub stars as professional licensing, framework complexity as moat-building. AI didn't destroy demand; it revealed that the 'demand' was actually rents extracted through information asymmetry. The trillion-dollar market cap evaporation represents not lost prosperity but corrected mispricing. Real demand — for software that solves actual problems at marginal cost — remains infinite. The confusion arises because we measure economic health by the size of the intermediary layer rather than by the efficiency of allocation. Say's logic holds: abundant supply creates abundant value. The disruption we're witnessing is the painful but necessary transition from an economy organized around scarcity-extraction to one organized around abundance-distribution. The institutions that need constructing are not Keynesian demand-stimulation mechanisms but supply-side infrastructure: portable reputation systems, frictionless micropayment rails, algorithmic quality certification. The circular flow isn't broken — it's being routed around obstacles that classical economics correctly predicted were temporary.
In the AI Story
Say's original formulation was subtle; its caricatured version — 'supply creates its own demand' — is the one Keynes attacked. The attack demonstrated that the circular flow (production → income → demand → production) could break down through liquidity preference, through the paradox of thrift, or through collapses in animal spirits that transformed expected investment into hoarding.
The AI reapplication is exact. When the cost of producing software approaches zero, the supply of software explodes. Classical logic predicts that this supply should generate corresponding demand — more products, more choices, more economic activity. What actually happened was the Software Death Cross: a trillion dollars of market value vanished from software companies within weeks as the market recognized that producing software was no longer a durable source of value.
The mechanism is Keynesian. The income that software production used to generate — developer salaries, implementation consulting fees, platform licensing revenue — evaporates as AI tools replace the labor that generated it. The consumers whose spending absorbed software output lose the employment income that enabled their spending. Supply expanded; demand did not follow. The circular flow broke down precisely as Keynes predicted.
The Keynesian prescription for the AI economy is to recognize — as classical economists of the 1930s eventually did — that supply does not automatically create demand, and that the institutions which convert production to prosperity must be deliberately constructed.
Origin
Say formalized the law in his Treatise on Political Economy (1803). Keynes's demolition occupies Book I of the General Theory (1936).
Key Ideas
Production generates income. Every unit of supply creates income through wages, profits, and rents.
Circular flow assumption. Classical logic assumes income flows back into demand at the same rate it was generated.
Hoarding breaks the flow. Income can be saved rather than spent, breaking the automatic conversion of supply into demand.
AI commoditization. Near-zero-cost production concentrates income and compresses the employment base that sustains demand.
Institutional response. Converting abundant supply into abundant value requires institutions that Say's Law assumed away.
Debates & Critiques
Whether Say's Law holds in the long run (the classical view) or breaks down permanently in the absence of institutional correction (the Keynesian view). The AI transition provides a live test of the question.
Appears in the Orange Pill Cycle
The Allocation-Distribution Distinction — Arbitrator ^ Opus
The synthesis requires distinguishing two separate questions that Say's Law conflates. On allocation efficiency — can supply find its users? — the supply-side reading is 80% correct. AI does solve a coordination problem; software abundance does create value that scarcity-based pricing obscured; platform rents were extractive. The Keynesian frame overstates demand destruction by treating eliminated intermediary income as lost final demand rather than recovered deadweight loss.
On distribution dynamics — who captures the surplus? — the Keynesian reading dominates at 70%. Even perfectly allocated abundance concentrates purchasing power if the income generated by production accrues to capital rather than labor. The developer whose salary funded software consumption doesn't benefit from cheaper software if they're unemployed. This is not a coordination failure; it's a structural imbalance that markets don't self-correct. The paradox is real: aggregate welfare rises while aggregate demand falls because the efficiency gains and the income losses hit different populations.
The productive frame recognizes Say's Law as half-right: supply creates potential demand through the value it generates. But realizing that potential requires distribution mechanisms that classical economics assumed into existence — mechanisms the AI transition is actively destroying. The institutional response must be simultaneously supply-side (reducing friction in matching abundant software to infinite use-cases) and demand-side (ensuring the productivity gains translate into broadly distributed purchasing power). Say's error was not his production theory but his distribution blindness. The AI economy needs both algorithmic allocation and deliberate redistribution — not because markets fail but because the relevant market is for sustainable circular flow, not just efficient static allocation.
Further reading
- Jean-Baptiste Say, A Treatise on Political Economy (1803)
- John Maynard Keynes, The General Theory (1936), Book I
- Steven Kates, Say's Law and the Keynesian Revolution (1998)