I don't disagree.
The Anthropic / SF collaboration is a very weird one. SF isn't AWS. They get accelerated access to Claude / AI in their platform, which is important differentiator vs. Veeva (at least in Life Sciences). But what are they giving up? Anthropic will be able to train its models on all of that CRM data, and in the future, they will absolutely use that to compete against SF and other CRMs. They estimated a $30T TAM because they see themselves displacing other industries. Does it help SF today? Probably. Long run? Doubt it.
Anyone who's thought about this stuff in a corporate strategy setting and landed on one model / partner is simply a silly goose. Models are far too dynamic right now, and it's natural to expect different ones may excel at different things. That, coupled with the strategic and financial flexibility to insulate against token cost uncertainty, is how every corporation should be moving forward.
Google is always a player. Gemini is good and having cloud and GPU capacity makes them intriguing provider. They also have very good, well at least not crazy, leadership.