That is a good point; however, I'd treat it as a risk management. Instead of a long-term contract with a commitment, one side optimises their fiscal load by accepting the risk of an environmental or conditions change.
For those who think the LLMs are the greatest thing - _I_ cenrtantly assign myself to such a category. And I also see some correlation between Tim Berners-Lee's experiment with WWW and what is happening with all those LLMs (and what was built on top of or with the usage of LLMs). Not a direct 1-1, but in a part of adoption. It was new, it required adoption, and it was given (almost) for free. And it did work, and adoption is very high, and people would give up their favourite tool. When adoption grew enough, it came time to reap the benefits.
Wasn't it kinda the expected outcome?
All those subscription models without limits (requests are not a good limit) or direct connection to tokens as currency (use another word if you'd like to) are ROI negative. The era of heavily subsidized AI products can't remain forever.
They just need to change something to earn money, and tokens are a great way for inflation emulation. Every economist knows how to maintain inflation and emit money (tokens) when needed to turn to the new stage.
I remember how, in the early days, my friend was buying token packages instead of having a monthly subscription, and it did work well for him until the next model update, which required more tokens for the same operation, and after some time, the outcome for $10 yesterday wasn't the same as for today. So, the market figured out which currency it will use.
For those who think the LLMs are the greatest thing - _I_ cenrtantly assign myself to such a category. And I also see some correlation between Tim Berners-Lee's experiment with WWW and what is happening with all those LLMs (and what was built on top of or with the usage of LLMs). Not a direct 1-1, but in a part of adoption. It was new, it required adoption, and it was given (almost) for free. And it did work, and adoption is very high, and people would give up their favourite tool. When adoption grew enough, it came time to reap the benefits.