AI Perez: a Perez-framework theory of where AI value accrues as models commoditize

cpaik · x · 2026-09-16

Pace Capital launched "AI Perez," an interactive analysis applying Carlota Perez's tech-cycle framework to AI value accrual.

Core thesis: market size doesn't tell you who captures value; pricing power follows what customers can't easily replace.
Hypothesized shifts: as capable models get cheaper, pricing power moves from intelligence to deployment, then to customer context and ultimately distribution — leading model labs won't automatically control those businesses.
Compute demand may keep rising even if frontier progress slows, as more spend goes to validating outputs, running evals, and auditing agent actions — favoring different chips/systems than frontier training.
Orchestrating multiple models and harnesses can squeeze more useful work from existing models, so lower cost per task can coexist with higher total compute use.
Constraint: compute can be added fast, but getting permission to use customer data and earning trust to act on it takes far longer, defining later-stage bargaining power.

More from Venture