The frontier labs are currently at point where they are essentially fighting for smaller and smaller piece of the pie.
The amount of tasks open models are just fine for increases every month, so there is just less and less tasks where you need frontier models.
This achieves 2 things:
* companies that might have reservations about chinese models having training backdoor are getting tools to train its own
* the tricks deepseek uses to make building them faster can now be used by more companies, increasing competition even further.
So to answer the question: The pie is already shrinking, and that just makes it even faster
The amount of tasks open models are just fine for increases every month, so there is just less and less tasks where you need frontier models.
This achieves 2 things:
* companies that might have reservations about chinese models having training backdoor are getting tools to train its own * the tricks deepseek uses to make building them faster can now be used by more companies, increasing competition even further.
So to answer the question: The pie is already shrinking, and that just makes it even faster