Say's Three Categories of Demand
Say's framework distinguished three categories of demand, a taxonomy largely lost in the simplified reception of his work. The first category is demand that exists prior to any act of supply — the hunger for food, the need for shelter, desires biological and structural. The second is demand called into existence by a novel product — Say's own example was the pianoforte, a device that created the desire for a specific musical experience no one had previously articulated as a need. The third, almost entirely ignored by subsequent economic thought, is demand that exists prior to supply but cannot express itself because no adequate supply exists. This third category — latent demand, stored pressure, accumulated creative potential — is the category that explains the AI adoption curve with a precision no other framework matches.
In the AI Story
The distinction between the three categories matters because they produce categorically different adoption patterns. A category-one product — meeting existing, articulable demand — faces competition from existing solutions and diffuses through the familiar mechanisms of price and quality competition. A category-two product — creating a new market — follows the S-curve of innovation diffusion: slow at first, accelerating as the market is educated about its new desire, then decelerating as the population of potential adopters saturates.
A category-three product does not follow either curve. It follows what Say's framework predicts and the AI adoption curve confirms: a discharge curve, flat for a long time, then nearly vertical. The delay is not about awareness but about adequacy. The market has been aware of its need for years or decades. It has tried partial solutions and found them insufficient. When the adequate solution finally arrives, the pressure discharges through the new channel at a speed that reflects the depth and duration of the stored need.
The key mechanism is that partial solutions do not reduce the pressure — they increase it. Each generation of programming abstraction — assembly to C to Python to frameworks to cloud infrastructure — narrowed the translation cost without closing it. Having tasted the liberation of one abstraction layer, the builder felt the remaining gap more acutely, not less. The ratchet of raised expectations concentrates the stored demand rather than dissipating it. By 2024, the accumulation across forty-seven million developers worldwide — plus hundreds of millions of non-developers who could conceive of things they could not build — constituted potential energy on a scale no market survey could have detected.
The category Say identified but his interpreters forgot is the one that makes the AI moment economically legible. Conventional market research can measure expressed demand — purchases, subscriptions, willingness-to-pay for products described to respondents. It cannot measure demand that has never been expressed because the object of desire does not yet exist in a form respondents could recognize. The stored pressure was invisible to every instrument of economic measurement until the adequate supply arrived and the discharge began.
Origin
Say developed the taxonomy across the five editions of the Traité between 1803 and 1841, refining his distinction between different kinds of demand in response to criticism and his own evolving understanding. The pianoforte example appears in multiple editions and was Say's preferred illustration of how genuinely novel supply calls demand into existence. The third category — latent demand — is present but less systematically developed, and it is this underarticulation that made it easy for subsequent interpreters to collapse the three categories into a single undifferentiated concept.
Key Ideas
Three, not one. Demand is not a single phenomenon. The three categories operate through different mechanisms and produce different adoption patterns.
The third category is structurally invisible. Stored demand cannot be measured by instruments that measure expressed demand. The absence of transactions is coded as absence of need — an error that persists until the adequate supply arrives and the discharge makes the stored pressure observable.
Partial supply increases pressure. Each incremental improvement that narrows but does not close the gap between imagination and artifact raises expectations and concentrates the stored demand rather than releasing it.
Adoption speed measures category, not quality. The shape of the adoption curve is a diagnostic instrument: S-curve for category-two markets being created, discharge curve for category-three markets being released.
Debates & Critiques
Conventional marketing and economics texts continue to treat demand as a single phenomenon differentiated only by elasticity and preference, with no recognition that stored demand is categorically different from expressed demand. The taxonomy's absence from standard frameworks is one reason the AI adoption curve appeared anomalous rather than diagnostic.
Appears in the Orange Pill Cycle
Further reading
- Say, Jean-Baptiste. A Treatise on Political Economy, Book II, Chapter I (1803).
- Schumpeter, Joseph. History of Economic Analysis (1954), chapter on Say.
- Rogers, Everett. Diffusion of Innovations (1962).
- Christensen, Clayton. The Innovator's Dilemma (1997).