Username or EmailPassword
The article's emphasis was on 'personal computing' rather than every possible application of compact/low-power/special-purpose computing, so I can see why the author didn't get into embedded applications.
That said, you're very nearly onto an extremely significant point: the trend is definitely away from owning one big general-purpose box to owning a heterogeneous collection of small, specialised devices. Which starts to look more like the traditional embedded ecosystem in its general philosophy, only with one huge exception: ubiquitous integration. The real trick is going to be in getting all these user-oriented devices to integrate seamlessly and securely, so users can mix-n-match services and access their data from multiple devices at any time.
On a purely hardware level I can imagine, say, Apple producing a smart TV that can also do casual gaming a-la iPhone, waiting a couple years till that's well established and then releasing a snap-on box that boosts it up to full-blown console level, making it both a desirable 3D gaming platform just as the traditional console makers are falling asleep at the wheel again, and providing enough functionality to do general computing (e.g. run a copy of Word) as well, allowing it to do triple duty as an iMac-like PC as well. And neither device would require much internal storage, because users can either keep all their data in iCloud and/or on a local turn-key Apple 'iHub' NAS (a much more flexible successor to their rather old-fashioned Time Capsule back-up system). And then all this stuff is going to happily chat with your iPhones and iPads, and even MacBooks if you still bother to own those. And even that only scratches the surface of what might appear in future.
Essentially, personal computing is now entering a post-scarcity age... at least where hardware is concerned. Consumers can now afford to buy a specialised device for each class of tasks they regularly perform. Each device will still retain some general-purpose capability (e.g. you can type a letter on an iPad or smart TV; mostly it'll just be slower if you don't purchase a keyboard as well), but for its optimised purpose each one will really shine - certainly much brighter than the traditional general-purpose PC which does a bit of everything reasonably but nothing brilliantly.
The real challenge will be on the software side - getting every device talking to every other device with zero hassle and zero configuration/management costs for the user will be no small practical feat. The basic concepts needed already exist in isolation, but fusing them into a completely successful mass-market solution will be a non-trivial task.
Hopefully this is something the author will explore in future articles; looking forward to them already.
(Full disclosure: While not really a true nerd/geek, I do still keep my very first ZX81 up on my cupboard shelf.;)