From niche to necessity
The microeconomics behind expanding the boundaries of what humans can do
In a conversation with a friend we marveled at the short few steps to create a hyperrealistic video ad for a barbershop via a patchwork of AI products strung together.
Knowing which to use — better yet what is actually out there — was the unlock.
Product discovery is becoming harder. It will only grow.
Everything transitory—the knower and the known. — IV. 35
Then
Look down at your Smart TV remote.
Chances are there is a Netflix button.
An advertisement masquerading as a product.
For subscribers it is a useful product. A few less clicks to beam “Emily in Paris” into your visual cortex.
For others, it’s a constant reminder of your TV’s best complement. Netflix.
Smart companies always try to commoditize their product’s complement1.
Since 2011 —when the Netflix button was first introduced — TV prices have dropped 84% percent while the lowest cost ad-free subscription to Netflix has risen from $8 per month to $15.5 (or 94%). Cheaper TVs mean more subscribers and profits.
Many decry the fact the button is bad UI — I agree. But it’s been part of a masterclass in accelerating demand.
Now
Think about your smartphone. A pure digital screen. If you’re on iOS a consistent, polished experience. Android: freedom of choice.
You’ve heard about Apple Intelligence. And it’s delay2.
For the promise of AI3 to really break through it’s progenitors know it has to be more accessible.
In the U.S. 94.6% of households access the internet via their smartphones versus 73% via laptops.
To drive accessibility and accelerate the adoption curve we shift and move to the market that’s the biggest.
The battle is waging in this case4.
Apple’s goal: further commoditize the foundation model market.
OpenAI, Google, Anthropic’s goal: further commoditize the smartphone market.
Apple’s complete vertical integration gives it a huge advantage over TV manufactures of yore in the Netflix example.
And despite the ChatGPT integration deal struck, Apple’s Siri will serve as the omnipotent router for the completion of tasks.
No OpenAI button.
Future
There have been rumors OpenAI is working on a phone. Or at least a more “intuitive way to interact with AI”.
Perhaps a retina scanner.
Or as we know — that AI pin — Humane which has already flopped.
Hardware is hard as the argument goes.
But could you realistically imagine jettisoning your phone for a pin?
The iPhone was revolutionary because of its software and app store marketplace that democratized an app for every purpose.
The bet right now: generative AI will soon be the only software worth having.
The classic 2002 exposition of this by Joel Spolsky
Those pesky EU regulators require competitor products and services to function on it’s devices.
For simplicity’s sake, lets define the promise of AI as expanding the boundaries of what humans can do.
But there are others, too. Servers and chips are natural complements to LLMs.