Google's edge in persistent agents: distribution and data lock-in
An investor argues against the bear case on GOOG: models are cheap to switch, so data lock-in and distribution decide winners, and Google's Gmail, Chrome and Android give it the strongest base for a Muse-style agent.
I really appreciate the views of @PythiaR , but I actually come to the opposite conclusion on $GOOG. Models have ZERO switching cost. It's an API key. People gravitate to the perceived better model as soon as it's available. Switching costs come through data lock and distribution. And agent frameworks are very easy to replicate. Google has immaculate data lock and distribution. They have just whiffed on the model and the UX (so far with a tiny shiny moment in between). If I believe the model changes on a dime, and that agent assistants are easy, but that you can't replicate compute, data lock and distribution, I come to a different conclusion. Google's product vision is immaculate. They are integrated up and down and left and right and everything in between. The reference post says the game is shifting from who has the smartest model to who owns the persistent relationship with the user. If that's true, it should be Google’s game to lose. You could effectively argue that nobody owns more persistent user relationships. Particularly given GCP Gmail, chrome, youtube android etc.. a Muse type agent not a significant amount of work to launch for a company of Googles scale. The distribution and compute intensity is WAY harder than the product - which is actually laughably simple. I think search is challenged irreparably and that's the downside, but I love $GOOGL as an underdog and I take the other side of this trade. Model layer is overrated. Particularly in inning one of that evolution
This post quotes another post: x.com/pythiar/status/2113004129199243265