Token demand explosion

Cheaper tokens raise total token use far faster than prices fall, so total AI spend grows instead of shrinking.

Facts

Lane: Capital and talent
Wave: 3 of 12 (build order, not a date)
Needs: 4 boxes all the way back through hard links
Unlocks: 0 boxes downstream through hard links
Leverage: None: already built
Cluster: Around datacentre capex
Found by: cost-curve unlock lens

Evidence as of October 2026

OpenRouter processed 25 trillion tokens a week in May 2026, five times its volume six months earlier, while fixed-capability prices have fallen 9x-900x a year (Epoch, Mar 2025).

Around it

Needs · 3

Capability price deflationcapability · Built today
Hardware cost per tokencapability · Built today
Algorithmic efficiencycapability · Built today

Unlocks · 2

Datacentre capexcapability · Built today · helps, not required
Frontier lab revenuestage · Built today · helps, not required

Loops and off-ramps

Cheaper means morereinforcing loop
Taiwan chip disruptionoff-ramp · slowed

Openings researchers' judgement, October 2026 scouting

Crowding4 Lab overlap3.5 Small-team fit4.5 Buyer today4 Research openness2.5

Depth: light
Players: OpenRouter ($113M, May 2026); Portkey ($15M A, Feb 2026); LiteLLM; Helicone; Finout; Cast AI; labs' caching/price cuts
Wedge: Agent cost-per-task measurement and optimisation (not cost per token): attribute spend to completed outcomes and cut wasted agent loops for enterprises whose agent token bills are compounding. startup