Usage costs will come down with better hardware. Hardware is improving rapidly each generation.
But memory costs are going way up. And both OpenAI and Anthropic bumped up the price of their frontier models in April.
Supply will eventually catch up with demand. Then the prices will come back down.
Eventually either the supply will go up or companies will start buying fewer overpriced GPUs.
Either way, the price per token will come down as hardware improves and supply and demand reach equilibrium.