By now most people are aware that the cloud is a bunch of computers that, while not literally in the sky, might as well be. Indeed, most people think of the cloud as simply the ability to store your files and run services through the internet. But those who really care about technology know that it's more than that: the giant server farms run by the likes of Google, Amazon, and Apple also provide countless services to companies as well as individuals. And they're in a race to become master of the cloud.

But before we get into that, let's talk about how the cloud works. The most famous analogy comes from cloud pioneer and Amazon CEO Jeff Bezos. When electricity first appeared, a manufacturing company had to build its own power plant to run it. Utilities came along and allowed these same companies to buy electricity by the meter. It made sense for everyone, Bezos pointed out, since if you're a manufacturing company, you shouldn't be in the electricity business, and it shouldn't have to be your job to do it. Let an electricity company do it, and it'll do so much better. It's the same with companies and computing. Large companies need armies of IT staffers and large server farms to run databases and other software they need to do their job. But they're not in the software business, and they shouldn't be. That's the cloud: Instead of having to build your own power plant to run your factory — instead of having to build your own computer infrastructure to, say, sell insurance, or cars, or whatever — you just plug into the grid.

But in a way, this undersells the transformative impact of the cloud. The cloud is all that, but it is also a new paradigm. If you have a bunch of machines that users run on a network, should the software run locally on the machines, or on hidden servers, with the machines being simply "dumb clients," meaning just screens and interfaces, with the actual computing happening elsewhere? The answer to this will depend on technology and economics. In the 1960s, the answer was self-evident: Computers were so large and so expensive that you could only have one big one, and smaller mainframes on which people would do their work, which would actually be done by the bigger mainframe somewhere else. Personal computers flipped that around, and now a number of factors, including the internet and the declining cost of servers, is flipping that back around again.

In that sense, the cloud is a general paradigm shift in how computing is done.

The reason why this matters is because it enables us to see where the cloud is headed next. When computing platforms evolve, it goes like this: First, the platform needs to get built out. And then, vertical applications are layered on top. The earliest personal computers didn't actually have much software on them, except basic things like programming languages (what got Microsoft started). Then specific applications came on top. The first "killer app" — a piece of software that suddenly justifies the expense of a new computing platform — was VisiCalc, an early spreadsheet application. Companies had paper accounting worksheets, where calculations had to be painstakingly made and rechecked by hand, and fixing errors was a nightmare. With a PC, you could plug in the numbers, and watch the results magically appear. Fix one entry, and the entire thing recalculates itself. Magic! All of a sudden a PC stopped being a hobbyist toy and became a money-saving business tool. The world was never the same again.

Today, the cloud is basically a way to rent a virtual computer that lives on the internet. Since it is a new computing platform, we know the next step is applications on top of that that take advantage of what the cloud is.

I'm talking about AI.

Google, which trails behind Amazon on cloud computing, is hoping that combining its capable cloud offering with its AI expertise will reverse the tide. So if you upload photos to Google's cloud, it won't just store them for you, it'll tell you what's in them. (And, for example, if you're a social media app, filter out the nasty ones.) Other services include speech recognition, text analysis, translation, and so on. Of course, Amazon is not far behind with its own AI offering.

And it makes perfect sense. It's not just that machine learning is all the rage. It makes sense for the companies that want to use these services, since most of them can hardly build their own cloud and develop their own AI software. But it also makes even more sense for the companies that want to sell them. The more customers they have, the more data there is to train their algorithms, leading to a better service, leading to more consumers. This virtuous cycle, called a network effect, usually grinds competitors to dust. It might mean that only one company will become the master of the cloud.

And since the cloud is the new computing paradigm, that would make that company the master of computing. Place your bets.