How to start a startup
A lightning talk from Demuxed 2025
You really, really need these three things.
Text version
...I don't know what AI is going to do to the world.
The impact of AI is probably going to be somewhere between the impact of The Internet and the impact of Terminator.
I also don’t know what’s going to happen to job markets, but it could be disruptive. But in an AI world, I can think of at least one job that is going to be really hard for AI to displace: entrepreneurship, because someone needs to mobilize the robots.
Of course, being a founder has always been a risky job. Most startups fail. I want to talk about why most startups fail by sharing 3 things you need to succeed.
My credentials: I've started two companies that have hit some level of scale: Zencoder and Mux. Zencoder and Mux are also responsible in some way for creating VideoJS and Demuxed. Now I’ve personally done very little to make either VideoJS or Demuxed succeed, and so I’m going to give you my bonus 0th tip for starting a startup: choose good co-founders.
1: Start with a big problem (not cool tech). The best startup founders are engineers, but that leads to a trap. I talk to a lot of founders who imagine some cool tech and then want to start a company around it. Cool tech without a big problem never works, and a lot of startups fail for this reason. “A better way to do XYZ with video” is probably a bad startup idea. “A better way for customers to solve XYZ pain” is probably better. Find a big problem in a big market and then solve it.
Once you’ve found a problem, how do you solve it?
2: Have good taste. Taste is what separates the best founders from the pack, and it’s why engineers make good founders when building highly technical companies. Taste is also something AI doesn't have, and it's unclear if or when or how it will.
How do you develop taste? That's a much longer topic, but check out my favorite work of 20th century philosophy for the answer: Truth and Method by Gadamer. (Or ask Claude to summarize it for you.)
3: You need durable strategic power. This is sometimes called a moat. What does this mean? It means you need a structural reason why other companies can’t do what you’re doing. You can get to product/market fit without durable strategic power, but you can’t build a really big company.
Have you noticed how there are hundreds of companies in this industry that all kind of look the same? It’s hard to find a moat. It’s true in AI too, and I think most AI startups are probably going to fail for this same reason. There are only a few types of strategic power, so make sure you have one.
I love this industry. Most days, I love being a startup founder. But every day, it’s hard. If this is something you’re interested in and I can be helpful, reach out.