For those of us who have been in the computing industry for decades, it's natural to guess that the transition has been overblown, and tempting to predict continuing swings between centralization (in the form of cloud computing) and decentralization (in the form of more powerful PC's). After all, that's what we've seen in the past. But I don't think that's what will happen. My guess is that we finally have enough power and the right architectures, on both tiny devices and massive servers, to ensure that the swinging is mosty done, with future evolution tending to increase power both at the user's location and at the centralized servers.
That very power, however, is what is bringing the PC Era to an end. A PC is fundamentally a huge compromise -- powerful enough to do needed work, but small enough (originally) to fit next to a desk in an office. The PC shrank to fit under a desk, then on a desk, then in a laptop, all without requiring any fundamental change to the PC paradigm, in which workers were assumed to be more or less chained to their desks. But now the technology has gotten small enough to be used in ways never envisioned for PC's. Smartphones are the most obvious example, but there are also wearable computers like Google Glass, distributed sensors, powerful computers embedded in mobile robots, and ultimately body-implantable computers and "smart dust." None of these are scenarios conducive to the interface of a traditional PC, and all of them may be even harder for an enterprise to manage than PC's.
Worse still for the PC paradigm, these new devices are slowly undercutting the most basic assumptions of the PC world. As smartphones become better and better at recognizing speech and handwriting, how long will it be until a new generation expects similar functionality on anything like a PC? But once you have those things on a PC, do you really need a keyboard? But if you remove the keyboard, doesn't that make it a tablet? As Microsoft has inadvertently demonstrated, the user interface of a keyboard device like the PC and any touch screen device are radically and incompatibly different. No one thinks of their smartphone or tablet as a PC.
Increasingly, computing devices will have a variety of shapes and forms, sharing only a tendency to store and interact with long-term information on centralized services. This is why cloud computing isn't a fad, or even another pendulum swing. Cloud services can minimize the amount of maintenance and retained state on any kind of device, while allowing the re-professionalization of IT services on the back end, and allowing most companies to outsource the vast majority of their system administration tasks. Advancement in user interaction will take place on the devices, while applications will rely on increasingly sophisticated cloud-based services to perform virtually all non-interactive functions.
In short, all the services that require maintenance are moving to the cloud, while individuals are moving away from PC's to more specialized interaction devices.
Twenty years from now, when a child sees a PC in a computer museum, he will be flummoxed by the lack of a touch screen, the bulky keyboard and mouse, and the lack of speech or handwriting interaction. This, he will be told, is what they called a PC. And if he's taking notes, he'll do it by whispering into his phone, or sub-audibly verbalizing to his wearable computer, or perhaps by wiggling his fingers to manipulate a virtual keyboard only he can see. He'll be interacting with a computing unit built into to his clothing or implanted in his flesh -- a far more personal computer than any we've known to date, but a long way from what we've known as the PC. Ultimately, what we've known as the "Personal Computer" will be viewed as the first historical example of any type of personal computer -- and a long outdated type, at that.