If you haven't seen the 2019 video, you should definitely take a look at Microsoft's ideas about where we might be headed with computer interface design. It's bold, slick, and leaves out a really big question.
Watch the video, and ask yourself: "how much work are people getting done?"
Do you see anyone writing a document? Responding to an e-mail? Correcting the errors the automatic voice transcription software made in the meeting minutes? Writing software? Designing a building? Laying out a magazine? Precision work is almost totally absent from this vision.
And with good reason: glass keyboards are awkward! There's no haptic feedback anywhere in this vision! In fact, physical controls are almost entirely absent. That's great--I love soft interfaces--unless you want to, say, touch-type, align two elements with the arrow keys while watching the results, adjust the volume of your music player in your pocket, dial a number on your phone without looking, or play a video game.
In fact, I think there's a fundamental problem here which hasn't been addressed. Vertical workspaces are great for looking at, since they take up a large solid angle in your visual field when sitting. However, holding your arms up for 8 hours to interact with one of those displays is tiring; which is why it's nice to have control surfaces laid out horizontally. These two opposing goals mean it makes a lot of sense to maintain blindly-usable physical interfaces--so you can input data where it's comfortable, and read it where it's easiest to see.
It turns out that virtual controls make a lot of sense for some applications, especially those where high-bandwidth (either in rate or space) input isn't necessary. Look at the success of the iPhone for a great example. Microsoft is showing some terrific ideas here. It just isn't the end-all be-all of interface design, awesome as it looks.