S1: A $6 R1 competitor?

This argument that the data centers and all the GPUs will be useful even in the context of Deepseek doesn't add up... basically they showed that it's diminishing returns after a certain amount. And so far it didn't make OpenAI or Anthropic go faster, did it?

What is the source for the diminishing returns? I would like to read about it as I have only seen papers referring to the scaling law still applying.