You make several good points, but I was responding to "It doesn't make too much sense for Amazon to invest too much of their own money into AI."

They're investing their own money in AI, they're just consolidating that investment.

Beyond that, it requires some mind-reading. A "frontier model" that "automates everything it possibly can" seems to imply a scope beyond just robotics.

Also, even if their initial focus is on robotics, innovations in AI are highly likely to come from that area. It's a frontier that the major LLM companies seem to barely be thinking about, but is very important to truly integrated artificial intelligence. Some AI luminaries have argued that that kind of real-world interaction is an essential component of true intelligence. One of the biggest limitations frontier models have currently is the impedance involved in their interface to the real world.