What Becomes Scarce When Intelligence Becomes Cheap?

If AI ultimately turns intelligence itself from a scarce resource into a cheap, near-limitless factor of production, where do scarcity — and economic value with it — migrate? Who controls the new scarce resources, and who might capture the largest long-term gains from this round of the productivity revolution? Or who can profit from the migration itself, even though the window may be very short.

The Industrial Revolution amplified the value of machines, capital, and large-scale organization; the internet amplified entry points, data, platforms, and attention. AI may shift the distribution of scarcity again. As generation, analysis, and decision-making grow cheaper, five resources deserve attention:

Demand

Once productive capacity becomes abundant, the question shifts from what to produce to who needs it — and even to how demand is produced. This resembles the internet revolution; it may be merely an upgraded version of it rather than something new.

Trust

When content, products, agents, advice, and even digital identities can all be generated at scale, judging whom to trust grows ever more expensive. Brand, reputation, identity verification, track record, liability, guarantees, and insurance all stand to appreciate. AI will have its own identity system, and perhaps its own insurers and reinsurers.

Judgment and Selection

AI can produce in a minute what it once took a team a year to propose. What is genuinely scarce, then, is judging which proposal is worth doing. The cheaper Generation becomes, the more expensive Selection becomes.

Frontier Discovery

Answers are getting cheaper, and discovering genuinely new problems and capabilities worth solving is getting more important. This, of course, has been scarce since the beginning of humanity — until productivity improved it was merely an important-but-not-urgent matter. If AI delivers a leap in productivity, then new drugs, new materials, and the stars and the boundless sea may actually come within reach.

Physical Scarcity

Land, energy, natural resources, geography, infrastructure, licenses, top-tier manufacturing capacity, and human time cannot be replicated without limit — though this is not new scarcity.

Distribution and access may also remain scarce. I have not thought this through; perhaps there is also emotional value — much like the first point, demand — which again looks like an upgraded version of the internet revolution.

The most promising commercial opportunities should sit at the intersections of these scarce resources: demand × trust is the Personal Agent; trust × judgment is the Enterprise Agent or Decisioning Engine; demand × frontier discovery is the new technology company.

Conclusion

If the marginal cost of intelligence keeps falling, which scarce resources will the next economic system reorganize itself around? It is not enough to build yet another "AI + X" productivity tool. Whoever can define the new markets, the new property-rights structures, or the new mechanisms of exchange — so that these new scarce resources can, for the first time, be identified → recorded → quantified → priced → traded → compounded — may be doing what genuinely counts as innovation in the relations of production.