1) Why does the first one work but I get an error with the second one?

We assume you have only one MOXF (instrument). When you create the VST INSTRUMENT, as a "Rack Instrument", Cubase asks you if you want to create a "MIDI TRACK". That's the first Midi Track. If you wish to record more, create another MIDI TRACK. You can have one (MIDI Track) per each of the sixteen Part on this instance of the MOXF. You will notice Cubase is referencing this instance of the MOXF as "1". "1-MOXF6/MOXF8 VST".

In general, you want to use MIDI Tracks, not an Instrument Tracks. You can use multiple MIDI tracks with your MOXF if you are using SONG/PATTERN mode with 16 Parts. Each Part can be recorded to a separate MIDI Track.

Also, I'm still trying to understand what is the advantage of an instrument track versus a MIDI one? What "shortcuts" do you get when you use an instrument track, because there obviously must be some advantage to using them that I am missing.

An Instrument Track, typically, would be used for a keyboard synth that is not multi-timbral and does not have multiple outputs. Like an analog synth where you are addressing the instrument once as a whole. There are now exceptions to that rule now, as Steinberg has expanded the role of the individual Instrument.

But typically you want to setup the MOXF as a "Rack Instrument", using a different MIDI Track for each Part of your Song MIXING or Pattern MIXING setup. When the Editor graphic interface opens and that first MIDI Track is created, notice the MIDI OUT of that track is assigned to "1-MOXF6/MOXF VST - Midi In"

This returns the MIDI to the EDITOR's MIDI In. You should be able to play the keys of the MOXF and see the keys of graphic interface respond.
Create a new MIDI Track, it will automatically be assigned to this same MIDI OUT except the Out Channel will increment to MIDI Channel 2.

When you highlight this second MIDI Track assigned to "1-MOXF6/MOXF8 VST - MIDI In" Channel 2, notice
that PART 2 in the Editor is automatically selected. And that's how it works - when you select a Track in Cubase, the corresponding PART in the Editor is selected. You should work with Local Control = OFF, and allow Cubase to do all channelizing for you when recording MIDI data.

The MOXF is a multi-timbral (16 Parts). Use one MIDI Track per Part.
On the Editor's menu bar go to FILE > SETUP
Make sure you have selected the MOXF driver for MIDI Port 1, and Port 4 in the appropriate places.
Go to FILE > VSTi SETUP
Make sure you select your Focusrite driver for the Main L and Main R ports you have connected on that interface, this will ensure that audio "returns" to the VST Instrument's AUDIO LANE.

So in the VST Instrument folder you will see an AUTOMATION LANE, an AUDIO LANE (stereo, Meter has a left and right bar), then your MOXF MIDI TRACKS (a single bar meter, midi is not sound and therefore not stereo). During recording and setup, you can MUTE the Returning audio, by activating the "m" on the Audio Lane. This way you can monitor the MOXF direct. But this is your choice.

When you have create sixteen tracks, this first instance of the MOXF is full. If you wish to continue, you can do what is called FREEZE this instance of the MOXF VST and do what is called "unload the instrument". This creates a temporary (invisible) audio track of what you recorded thus far, it locks the MIDI TRACKS and plays back the temporary audio file (the Freeze file). This allows you to unburden the MOXF hardware from having to generate the audio data in real time.

Now you can open a second instance of the MOXF VST, it will be "2-MOXF6/MOXF8 VST", and it will be 16 Part multitimbral - free from having to reproduce the original tracks you can now build 16 more. That's the advantage of using the MOXF VST as a Rack Instrument with a MIDI Track for each Part. The frozen MIDI tracks can be kept in a folder and kept in case you change your mind. When you unFREEZE the tracks, you still have your original MIDI recordings so you can still make corrections and/or changes.

VST routing of your external hardware allows you every benefit afford a soft synth VSTi that is generated entirely on the computer. This includes Freeze, includes processing returning audio through plugin effects, includes EXPORT AUDIO MIXDOWN... the only thing is when you render audio via "mixdown" it is real audio so it must be done in real time.

Hope that helps.

Thanks so much Badmister!

Just to make sure I got this routing straight:

I have my line outputs (L&R) from my MOXF plugged into my Focusrite inputs 3 and 4 respectively on the back of the unit.

Cubase triggers the tone generator with the MIDI performance, then it goes out the tone generator to the line outputs which are plugged into the Focusrite inputs 3&4. This is where I guess the MOXF VST taps the audio from the focusrite based on this setting?

So this "audio lane" is basically the line outputs?

I cant seem to edit my response above but the picture may be hard to see. The first two slots on the VSTi setup are input 3 and 4, and the next two are unassigned.

Also, it seems both MIDI tracks are mixed to one audio track (MOXF6/MOFX8 VST). Is there a way to isolate the audio tracks?

That's correct, Using the audio returns 3&4, if that is selecting the Focusrite's driver.
In the MOXF you will be using "1StereoRec": this translates to Single Stereo Out for Record.

You don't get to use "2StereoRec" because you've only connect the Main L/R output to your audio interface.
The second pair of outputs is only available to those using the MOXF as their audio interface via USB. Via USB the MOXF's audio interface is Dual Stereo Out for Record. (USB 1/2 and USB 3/4)