Subsidise now, profit later. If you have the cash, it’s probably better to burn some getting early user adoption if you believe the cost base will reduce in th

Subsidise now, profit later. If you have the cash, it’s probably better to burn some getting early user adoption if you believe the cost base will reduce in the future. Inference costs are dropping fast, so you might as well subsidise now and get market share if you think future costs will drop below future revenue. User aquisition will be much harder later when you lose first mover advantage.