Because when inference providers user/learn from this engine, their prices will go down, causing commercial API pricing from closed source providers to go down, hurting their net revenue.