Are LLMs an inflection point?
#10 | On the future of software
This edition of Unbeaten Path was edited by my friend Chris, who skipped the music this time while polishing my words, as he was with me in Italy!
In evaluating a startup, the "Why Now?" is of incredible importance. If analyzed in enough depth, the answer to this question leads us towards how the world will look in a decade from now.
The first decade of the 21st century is a good example: By 2003, more than 50% of the US had internet access, and the iPhone was released four years later. These two events set the basis for the feasibility of a very powerful business model: the “closed ecosystem” business model based on apps.
Still, even today, Apple charges a 30% fee for payments made on iOS apps.
The same events also set the basis for the death of what used to be very powerful business models. Take newspapers as an example: A summary of the news was a damn good business, until it wasn't anymore. The Internet arrived, changing how we consume information and the speed at which information travels shifted - hello Facebook. Former empires rapidly collapsed.
Now that you can have real-time information (internet), delivered to your pocket (iPhone), ranked according to your preferences (Facebook’s algorithms), why would you buy a piece of paper that's already obsolete before lunch?
Unfortunately, looking forward instead of backwards, we don't have the benefit of hindsight.
But let's try to connect the dots.
App stores work because there are many people who can build apps, and there are many more who need an app to do something. However, these two subsets don't often overlap. This creates a marketplace, and whoever owns the marketplace, makes the rules.
Yet, when one of supply or demand stops to exist, marketplaces collapse. To me, it's unreasonable to think that the subset of people who need an app to do something will shrink in the foreseeable future. On the other hand, I find it completely plausible that the intersection of those who need something, and those who can build it, will dramatically increase.
Picture this: your flight is late, and you want to play a little game on your phone to waste some time. You can't code, so there's no chance you can build that game on your own. Luckily, someone else can: you open the app store, search for “Flappy Bird”, click download, and voila - the magic of meeting supply and demand.
The fact is, now you don't need to know how to code to build a little game. You just ask your favorite LLM (I like Claude) and in a few seconds you have your game on the phone. Don't trust my words, trust your eyes; the following is a recording of my screen I just took while writing this:
Crazy, right? So what are the implications?
Hold on! I'm not saying Apple will collapse. But it's not unreasonable to think that closed ecosystems like iOS will look very different in the future. Why do we need an app store, when we can create an app in a matter of seconds?
There are exceptions of course: Services that require many people to be in the same place at the same time - think marketplaces such as Uber, or social networks like Facebook - these, I believe, will still rely on a mobile or web app to connect people. However, closed ecosystems might lose some of their power in the future, and this is very desirable for those that didn’t benefit from closed ecosystem but were forced to use them because of their strength.
The most notable case of a business that hates the closed ecosystem game is Facebook. They hate it because they don’t own it, hence the Metaverse, which was their attempt to own the next generation of closed ecosystems. With some speculation, we could consider that Zuck's focus on LLMs is his attempt to nail the lid tightly on the closed ecosystem coffin, and to allow open ecosystems to run riot.
Speculations allow us to imagine different futures: when an inflection point comes (the famous “Why Now?”), it generates a second order of inflection points - more or less direct consequences, that change markets and create the foundations for big businesses.
What happens if our speculation above turns out to be true? How does consumer software change? And what about the hardware needed to sustain new needs?
I don’t know the answer, but I’m pretty sure these are some important questions to ask.