Balaji Srinivasan explains the distinctive features that make Internet startups attractive and investable. He lists and illustrates several core properties — operational scalability, large addressable markets, software generality, low capital and regulatory barriers, open source foundations, long-tail markets, failure tolerance, and the value of exposing business logic via APIs — and shows how these features let software hollow out traditional industries and be integrated into existing physical businesses.
Chapters
What are the key features of Internet startups? So one of them is something that we'll call operational scalability. So it's now feasible for you in your room to trade with someone across the globe at any time of day without special hardware. You don't even actually need to be at your computer to do this. You can just set up a store and take a transaction from them or from 100 people at the same time. We don't really think about how amazing this is, but think about how different that is from the physical world where you need a clerk and you have lines and queues and things like that. An Internet store can operate 24, 7. There's no zoning regulations, there's no limitations basically in terms of how many stores you can have in a particular area. It's something where the scalability is absolutely crazy. You don't need clerks, you just need databases and the ability to handle the transactions. So second, and by the way, when you think about the fact that Amazon's website gets a lot of traffic, just imagine how much their back end gets in terms of traffic. If the website gets a lot of traffic and that can operate and process transactions within microseconds or less, then at some point those transactions are being handed off to a physical operation which is definitely not processing them in microseconds. It's much, much larger. So this gives you a sense of why operational scalability is so interesting. All these things that were previously physical and constrained are no longer constrained in the same way. A second major property is market size. So with an Internet startup, in theory, you can go after the entire world. You can more easily attack a market in Indonesia or South America than you ever could before. Friendster, for example, famously became very popular in Indonesia, more so than in the US for a long time. And so even within a country, a business in North Dakota can markets where it's just the entire country, it's no longer constrained by geography or protected by it. So things become. There's not just more opportunity, there's more competitors. Everybody is in some sense adjacent to you now. And you have to basically win a winner take all competition to win your particular niche. Generality. So software is in many ways the most general product imaginable. And an Internet startup is delivering software anywhere. So you know, whether that is ideas like news and blogs, entertainment, material goods, communications, transportation, medicine, anything, anything you can think of, software is upstream of it. And you should read this article, it's one of the few articles I put at the very beginning of the class. Software is Eating the world. It's a very good piece that encapsulates what is obvious to people in the Valley, but is not yet obvious, I'd say, to the rest of the world. Anything that has an informational component, which is pretty much every industry, is going to be hollowed out with a database and an API at the center of it. And the company that does that is going to have operational efficiencies that are far greater than any of the existing incumbents. And that's what's happening right now with Uber doing that for taxis and Airbnb doing it for hotels. And I'd say that our company council is also doing that to some extent in laboratories and for genomics. And so the generality of software is why it's feasible for a company like Google to go from a search engine into maps or Amazon to go from a bookstore to a server farm. You know, software engineering skills are extremely portable and more so than probably anything besides pure math. So let's talk about low capital barriers. Basically, you know, you can kind of think about that hyper deflation of hardware costs that was begun by Moore's Law that just keeps going. Yes, it's true that Moore's Law has run into some heat considerations, so it's going parallel. But the fundamental aspect of transistor density doubling and effective computer cost declining is still true. And so for a few hundred dollars, you can purchase this super powerful general purpose computer that you can use to develop arbitrary software and communicate with any other computer in the world. And a useful way to think about this is, you know, if you're a mechanic and you need a special wrench, right, you need to go to the, you know, store, or you need to go to the catalog and order that wrench. If you're a computer scientist or a software engineer and you need a special tool, all you need to do is think it up and then type furiously into your keyboard. And suddenly this thing, you know, this wrench of pure electrons is basically there. And then you can use it to tweak whatever piece you want. You can think of it as your own 3D printer before the term 3D printer. So capital is very. Capital costs are very low because you can often build your own tools. Regulatory barriers are likewise low. Basically, the thing about the Internet is it's difficult to regulate. A discussion forum like Hacker News could pull in assets from servers in three countries and comments from a dozen more and make it very difficult to determine jurisdiction. But this is something where we have to keep vigilant. This is not Assured of remaining true forever. Open source is another key feature. The thing about Open Source is that the modern Internet is built on open source. This is a great essay on how open source economics are distinct from both communism and market economics. It's not a matter of left or right. It's really kind of orthogonal to these axes by somebody named Steven Johnson, who's a great author in the US So it's worth taking a look at. And not just open source. Another key feature is a long tail. With the Internet, you can go after micro targets. You can go after a market that is maybe 10,000, 20,000 people worldwide, whereas before, your cost of customer acquisition, of finding and acquiring those customers, let alone building a product for them, would be too high. Today, you can go after what we're, you know, what's called a long tail. That is to say, you've got to visualize a tail where your bestsellers or your biggest, you know, markets are over here. And then a tail going out all the way over here to here are the people who really like T shirts with basketballs on them or some kind of micro market. And then you can go and target that market, because with search, you can find those people who are actually searching for T shirts with pictures of basketballs on them. And then you can actually, you know, serve that market. And, you know, paraphrasing, you know, Winston Churchill, never in the history of mankind have so many earned so much from so few from such small markets, or learned so much. You can also have that because now we have online education or MOOCs like this course. So another example is failure tolerance or another kind of property. When Google or Facebook crashes, people yawn and wait for the site to come back up. But when there's a physical car that crashes or drug is contaminated, people die and there's new regulations that are passed. So this is a huge difference, a colossal difference in the penalty for failure. And it's one of the big reasons why Internet starts are allowed to move fast. You can move fast and break things in a way that you just can't in the physical world. But crucially, because we can simulate more and more of the world on the computer via CAD cam and virtual wind tunnels, Monte Carlo risk models and stress tests. Dying can be done on the computer many, many times before we actually die in real life, before we hit enter and ship the final product. It's kind of like playing a video game. And you can simulate, simulate, simulate, simulate, and thereby de risk the approach in a way that a traditional physical competitor would never think to do. Finally, one of the perhaps most interesting aspects is that you can titrate these distinctive features of Internet startups into existing businesses. One of the best ways to do that is to build an API such that all your business logic, all your orders, all your things that are core to the business are computerized. And then you interface with the physical world as little as possible. For example, in our laboratory, we have physical interfaces where you actually take a test tube and then you run through a machine, and then it's all bytes after that point, then you're on the command line. Then it's a pure software operation. And the goal is to get onto the command line, to get onto the Internet as quickly as possible with as much automation in that physical interface as possible. And if you do so, then now you're an Internet startup. Now, the key is, of course, any physical layer that you add is going to introduce friction. It's going to choose cost, and it's going to become a very important part of your business. But if you think of it as almost like a device driver, in the same way that you hit enter and the software will tell your hard drive to move the platter around somewhere, or you hit enter and it will turn on a light or not, you, you think of that physical layer as an actuator and it's controlled by software. And that's a very, very useful way to think about how you should do business going forward. So those are some of the key features of intranet startups. These work together synergistically. They're one of the reasons that they're very attractive for large investments. You should read this paragraph and whatnot.
