Yes, those are scaling laws, but when we see vendors improving their models without increasing model size or training longer, they don't apply. There are apparently other ways to improve performance and we don't know the laws for those.

(Sometimes people track the learning curve for an industry in other ways, though.)