Is Technology a Moat?
Content is king, always has been
Recently there was a lot of fanfare about how to develop a moat in AI if everyone depends on the SOTA models to begin with. Reasoning goes, you don’t own the weights, so anyone can come along and fine tune them with superior data or use a later, higher parameter model and eat your lunch.
Furthermore, what is the point of developing a service or product using OpenAI’s platform if they were just going to gobble it up later?
What is stopping OpenAI from pulling an Amazon and using the data from third party sellers on the platform to launch their own competitors?
Developing on a platform always binds you to the whims of the platform owner…
Speaking of platform owners, I was recently working on a project for quick deployment of infrastructure - just run the container build command for the logic, a deploy command and 🫰it will be distributed globally to a container repository, where it can be invoked upon request of that service from any arbitrary endpoint.
DHH and his team open sourced a tool earlier today which does that and makes it reachable 😢 called MRSK - it looks fantastic, I plan to use it for myself. I think that DHH ships great stuff and I am a big fan of the whole team at Basecamp.
Related - I was recently *wowed* big time by Cloudflare Pages - a JAMstack solution to deploying web applications in a quick and headache free manner. Something that made me consider if I needed to roll my own deployment software.
There is a ton of competition heating up to win serverless deployment - and for developers it will provide a lot of opportunity to disrupt other companies and industries by providing services at a much cheaper price.
Setting aside my shattered dreams of making a dent in the universe with respect to deploying a MRSK-like tool - MRSK, Moats, and another paper I read recently called “Commoditize your Complement” by Gwern got me thinking: is technology a moat at all? If it isn’t what *IS* a moat?
Commoditize your complement shares an idea first popularized by Joel Spolsky in “Strategy Letter V” - where he states that the reason enterprises fund and contribute to Open Source software is to cause deflationary pricing pressures on other parts of the stack. This puts pressure on near-field competitors while also increasing consumption of complementary goods (like the contributor companies’ core product offering).
So - if technology enables business innovation, and that business innovation hopefully enables disruption, but you don’t win a monopoly - what happens?
If the disruptive technology was a moat, then we could expect a few things to be true:
New firms won’t be able to compete in the same layer of the overall use case because they can’t recreate your innovation (until your patent runs out) → Incumbents suffer mounting losses and the innovator wins out in the end
Instead we can look at a few recent examples that until recently were held up as the golden-children of disruption:
Uber - using smartphones to hail cabs and defining drivers as contractors they build a system that has proprietary software and algorithms. The smart phone disrupts the taxi industry. Today - Lyft competes with them and they fight an eternal pyrrhic war over who can sell the most dollars for 50c.
Netflix - using mailed disks and then streaming, they disrupt the movie rental business and the cable television business. They more or less innovate high quality video streaming. Today - Netflix can barely make ends meet and everyone and their cousin has a software solution to stream high-fidelity content.
There are more examples, but these two should illustrate the core of what I want to share, Netflix more than Uber - and that is: technology is not a moat. It is a cannon, a head start, a handicap, heck it may even be a bog (which is like a moat) - but sadly reader it is not a moat.
Eventually other smart engineers find a new way to do that thing without infringing on your IP and then you have a competitor who can more or less do the same thing as you.
Scale isn’t even a moat!
So what is?
Content and a brand. They are the only meaningful moats around. And I think that this ultimately means that any product or service that can’t fit itself into a larger social or story narrative is doomed to fail.
When someone won’t drink coffee unless it’s Starbucks because they wouldn’t be seen dead with a dunkin’ cup - that is a moat.
When someone won’t think of using another plumber because their plumber is so professional and fairly priced - that is a moat.
When someone has tea instead of a cola beverage because they don’t have Coke, only Pepsi - that would be a moat if it ever happened.
Mickey Mouse is a moat. Captain America is a Moat. The Office is a moat. Apple devices are a moat.
They simply can’t be replicated - only imitated and those imitations are known to be such unless you have a virgin audience, or you get in early with the next generation.
Technology has always been a better disruption engine than a moat but I have to say - scale can be a moat. Look at Netflix's PnL statement today. It's barely making any money but as free money disappears, it's the only one that has a sustainable path to keeping cost of goods stable (content investment) while increasing revenue thus driving enormous income growth via operating leverage. A new entrant cannot compete on scale unless they are setting money on fire.