Smarter machines are bad for supply side moats
In the mid-2010s, I remember thinking that the biggest threat to Uber was the possible arrival of self-driving cars. I figured that if self-driving cars replaced human drivers, then Uber’s network of drivers would no longer be a strategic advantage over its competitors. With large enough pockets, a competitor could enter any given market by purchasing a fleet of autonomous vehicles.
I’ve been thinking this week that the self-driving question for Uber is actually an instance of a pattern that we may see more of in the coming years. The pattern is that machine intelligence eliminates many instances of supply side moats.
A few quick examples:
Social graph vs algorithmic feeds.
One way of thinking about the moat here is that audiences spend their time in social apps that connect them with the content creators they care most about. For a time, the most popular social apps were those that connected users to content created by their friends.
But with TikTok, Instagram Reels etc…, it’s become apparent that users are willing to spend their attention in social apps whose value proposition is not seeing content from friends, but seeing content that is personalized to their tastes.
Basically, traditional social apps are finding that the social graph is an ineffective moat against hyper-personalized content providers….
UGC vs machine-generated content.
Like social networks, UGC networks are two-sided: contributors post because they want recognition from consumers; consumers spend their attention on these networks because they are interested in the content that is being produced.
We can think of penetration among content creators as a kind of supply-side moat.
But in a world where machines can generate high quality content, existing networks are no longer reliably the best content providers. For instance, in increasing cases, ChatGPT is actually superior to Stack Overflow as a provider of code-related content.
Premium media vs synthetic media.
Spotify and Netflix sell access to premium content.
The dynamics of their business are certainly different in important ways, but their moat has the same effect: for competitors to beat them, they have to offer users access to a preferred catalogue of content.
If we get to a point where services can generate synthetic content that is personalized to our tastes (I can squint and imagine a Spotify killer that generates personalized music), then premium media providers will find themselves in a vulnerable position.