The 15" Retina Macbook Pro (RMBP) is the first "perfect" PC. Of course Apple and its suppliers will continue to innovate, but the innovations at this point are effectively meaningless because the RMBP has extracted virtually all the value to be had from the laptop/PC form factor:

It's super thin and light. It could get thinner, but I can comfortably carry the RMBP all day without any problems.
It's powered by the superfluously fast quad-core laptop grade CPUs Intel offers.
It can support up to 16GB of RAM, allowing for nearly limitless multi-tasking.
The logic-board integrated SSD based storage is ridiculously fast. The OS boots up in seconds, and all applications and content load instantly.
The retina-grade screen is stunning.
The trackpad with BetterTouchTool is spectacular.
The keyboard with Alfred is incredible.
It offers native HDMI-out (finally!).
It runs the sleek Mac OSX, with options to virtualize any other OS via Parallels Desktop or VMWare Fusion. Modern virtual machines offer performance that rivals that of native installations.
It can power 9M pixels across 3 monitors without any problems.
The battery is good enough for all of my mobile requirements.
All of my content seamlessly syncs to all my devices via iCloud, DropBox, and Gmail/Google Drive.

I use my RMBP all day everyday across 9M pixels and 3 monitors, and I'm always running at least 30 apps. It never slows down under any circumstances. All applications run at full speed at all times (I don't count in-app bugs against the hardware + OS experience).

What more could I ask for? And what comes next?

To answer that question, let's revisit the most basic question of computing: what do computers do? They take input, they process it, perhaps store it, and then may output something that the user can perceive through at least one of the five senses. Note that the processing and storage don't have be done locally, but could be done in the cloud. We have fundamental opportunities to innovate at the HCI input layer, the hardware performance layer, the software layer, and the HCI output layer.

The RMBP keyboard and trackpad are phenomenal. They're efficient, precise, accurate, comfortable, and elegant. There's not a whole lot of innovation left to be had on the keyboard and trackpad. Microsoft's hardware partners have been offering all-in-one touch screen computers for years, but they haven't been very popular because desktop based touch screens aren't ergonomic. Windows 8's lackluster critical and commercial reception speak to the challenges of making effective use of touch-screen technologies in traditional mouse/trackpad/keyboard form factors. OSX supports global voice dictation, but it's not nearly as useful in the PC form factor as it is in the smartphone form factor. There are opportunities to integrate new HCI mechanisms such as Leap-like technology into PCs, but the use cases for Leap's technology are inherently marginal. Leap presents opportunities for novel apps that may not have been feasible on a keyboard/trackpad, but Leap won't change how we use laptops or desktops. There are still other opportunities to innovate at the HCI layer, such as eye tracking, but it's a decidedly uphill battle. Most of the technology press found the Samsung Galaxy S4's eye tracking to be a marketing gimmick and not particularly useful. It's going to be extremely difficult to meaningfully innovate at the HCI layer in the laptop form factor. Many people have tried and failed.

As I outlined above, the hardware performance layer has passed the point of relevance. Even "breakthrough" technologies such as memresistors will provide marginal performance improvements given how fluid the RMBP is today. I am still waiting for that always-promised-but-never-delivered revolution in battery technology, but the benefits of battery breakthroughs will be far more profound in hyper mobile form factors such as smartphones, watches, and eyeware computers, and in transportation such as cars and planes. Battery breakthroughs won't really change how we use laptops.

We've been iterating desktop OSes for 30 years. Desktop OSes matured some time ago. They are optimized for the HCI input and output mechanisms: the keyboard, trackpad camera, and microphone; and the screen and speakers. If the HCI mechanisms don't change, there's no reason to make any dramatic changes to the OS. Beyond HCI optimization, the dearth of ground breaking new features in PC OSes speaks nice to lack of innovation opportunities left. It was nice to see Apple integrate iCloud directly into OSX to enable seamless file syncing across devices, and to optimize OSX for retina grade displays, but there's just not much left to be done at the desktop software layer. Desktop OSes are gorgeous, functional, always synced to the cloud, are very optimized for their HCI mechanisms, and lightning fast.

We don't have a lot of innovating left to do on desktop computing, other than to reduce the cost of perfection. The 13" RMBP starts at $1500, and the 15" at $2200. Perfection certainly isn't cheap, yet.