Collapsing inference prices turn compute giveaways into the customer-acquisition budget.
With each model generation shipping step-function price cuts and open-weights rivals undercutting closed labs, AI vendors can now buy adoption with subsidized tokens and credit drops the way cloud vendors once bought pilots with compute credits.
The same conclusion keeps arriving from across the workspace's research — 1 topics independently instantiate this theme. Filter the evidence by where it came from:
An order-of-magnitude drop in token prices lets usage-based products subsidize customer acquisition with inference while still holding margins.
Open-weights entrants weaponize free, modifiable weights and half-price launch windows as a pricing wedge to buy adoption away from closed incumbents.
Launch-week credit gifting is booked as customer-acquisition investment rather than operating expense — the accounting signature of compute-as-CAC.