The Market Learns to Price Intelligence

Markets are good at pricing visible scarcity. They know how to value oil, bandwidth, labour hours, chips and warehouse space. AI asks them to price something stranger: the industrialisation of cognitive work. If intelligence becomes cheaper, faster and more abundant, almost every business model becomes a question mark.

From labour saving to judgement saving

Markets have priced machines before: looms, railways, turbines, mainframes, personal computers, smartphones and cloud servers. AI is different because it prices not only speed or scale but judgement-like work - writing, coding, classifying, advising, searching and negotiating.

The cautionary precedent is the productivity paradox. Robert Solow joked in 1987 that computers were visible "everywhere but in the productivity statistics". The point was not that computers were useless. It was that organisational change, complementary investment and measurement lag behind invention.

Computers appeared "everywhere but in the productivity statistics".

AI markets may repeat that pattern at higher speed. Share prices can capitalise a story before balance sheets prove it. The durable value will come not from replacing workers in a spreadsheet, but from redesigning workflows so that cheaper intelligence becomes safer, faster and auditable.

The optimistic case is straightforward. More intelligence lowers the cost of research, coding, design, administration, customer support and scientific discovery. Firms become more productive. New companies appear. Old bottlenecks loosen. The market rewards those who turn models into reliable workflows.

But markets can also overprice a story before institutions catch up. A company may announce AI adoption without real productivity. A platform may spend heavily on compute before demand is proven. A labour market may reorganise faster than training systems can respond. A model may look cheap until energy, inference, compliance and liability are included.

The deeper market question is who captures the surplus. If AI raises productivity, do workers gain wages, consumers gain lower prices, shareholders gain margins, or platforms gain rents? The answer will vary by sector, but it will determine whether AI feels like progress or extraction.

Markets will learn to price intelligence through mistakes. Some firms will buy theatre. Some will build durable advantage. The difference will be operational discipline: data access, workflow redesign, governance, measurement and a willingness to change incentives rather than merely add a chatbot.