Turning point · Current state: Concentrated

Do the gains spread or concentrate?

This tracks whether AI capability, ownership, profits, infrastructure, and decision power diffuse broadly or accumulate among a small number of firms and governments.

Why this matters

A technology can be widely used while the power behind it stays concentrated. This turning point determines who can shape the systems, capture the largest rents, and set the terms for everyone building on top of them.

What the states mean

Lower endBroadly shared

Capabilities, ownership, and rents diffuse across many actors; switching and entry are relatively easy.

MiddleMixed / contested

Use diffuses broadly while strategic control and economic rents remain contested.

Higher endConcentrated

A small group retains durable control of frontier models, compute, distribution, and a disproportionate share of value.

What we look for

Concentration in frontier model developmentTracked as evidence about this turning point.
Compute, cloud, and chip market concentrationTracked as evidence about this turning point.
Open-model performance relative to closed frontier systemsTracked as evidence about this turning point.
Switching costs and portabilityTracked as evidence about this turning point.
Distribution of AI-related profits and rentsTracked as evidence about this turning point.
Access to training compute and capitalTracked as evidence about this turning point.

Indicators underneath this question

  • Compute / cloud / chip concentration
  • Frontier-model organizational concentration
  • Competitive diffusion / open-model counterpressure

Futures this can move

  • Augmented Society
  • Concentrated Power
  • Broad Abundance

What would move the tracker?

The state changes when admitted evidence moves the underlying indicators across the frozen scoring rules. The model should not move because a story is prominent in the news; it moves when the measured condition changes.

See evidence events over time →