Building or buying prebuilt systems

- [Narrator] Now we're tackling a fundamental question that everybody asks themselves, and that is should you build your own system or invest in a pre-built solution? So let's start with pre-built systems, right? There are many companies out there that offer workstations, desktops and servers that are specifically designed for AI workloads. These systems come with many different advantages. You know, first, they're tested and validated. The components are guaranteed to work together. And the systems, in some cases, are benchmark for machine learning task. And there's a few companies that are now starting to specialize just in AI systems. Now, the challenge with that is that they are very, very expensive right? Now, the other thing is pre-built systems do have a few additional drawbacks. Typically, they come with a premium. Of course, you pay a lot more than building your own system in many cases. And you're also limited to the configurations they offer, which may not perfectly match your need. Specifically, again, you are the owner of that lab, you're the owner of the scope. You're the one that actually wants to know what type of models you're going to be running, what type of learning you're going to be doing or what type of things you're going to be putting, quote, unquote, in production within your home environment, right? At the end of the day, in some cases, it's not so much of just learning, but also taking advantage of these newer models that can automate a lot of things, you know, within your environment, right? Now let's explore the DIY approach, which is building your systems from scratch. We just went over, you know, a few notes related to hardware and what type of hardware you should be using, right? But one of the most obvious advantages in building your own system is, of course, cost savings, right? By introducing components directly, you can often build an equivalent system, or even better for significantly less money. You also gain complete control over every single component choice, and allows you to optimize your specific workloads. Building your own system also provides a really good knowledge, right, that you'll understand your hardware intimately, making troubleshooting and upgrades easier, you know, down the line. Plus, you can customize aspects like the cooling solutions, cable management, of course the GPUs, the CPUs and memory and so on. But building isn't without any challenges. You need to research component compatibility, handle system integration, you know, yourself, manage multiple warranty claims if issues arise, and, of course, tackle your own troubleshooting. There are some other specific scenarios where each approach will shine, right? The prebuilt systems make sense whenever you need systems up and running very quickly, you don't have any, you know, worry about the cost, and you have limited technical expertise in-house to actually build a system, right? Now, build your own is often better whenever you're working on a tight budget or you need highly specific configurations and you actually have the technical expertise available. And at the end of the day, if you built any other PC for gaming and so on, it's basically exactly the same approach in here. In the next section, we're going to go over a few tips on choosing the operating system, whether you're going to be running this on Linux, Windows or macOS.