Startups are literally between rock and a hard place in this AI world.
Rock = the Googles and the Microsofts that develop powerful general AI models that work as a “thick protocol layer” leaving little room for the applications layer. There’s a reason why many of AI apps we’re seeing are wrapper apps with some clever UI elements for accessing the same models
A hard place = legacy industry players that haven’t fully embraced AI yet, but are sitting on massive amount of historical user data, such that when they flip on their AI switch, startups that were trying to build and fine tune vertical AI models will immediately have tough competition
Then, what can startups do? Two possible moats for AI startups:
Network effect: If you can build a platform where more users join, the overall value of the network grows exponentially, such a flywheel is very hard to turn off and eventually becomes an insurmountable moat. This still holds true in this AI era
AI-first product approach: Winners from the mobile era were those that rethought product experience from a mobile standpoint, not those that shoehorned the same web experience into mobile. What are some of the unique new things that are only possible now because of AI (as opposed to, AI making existing products or processes better)? If AI was invested first, then mobility (or insert whatever else) was invented after the AI, what would mobility look like?
Better yet would be combining the two – for example, a completely new AI-enabled content platform that, over time, achieves a strong network effect.