Another consideration is that if you fall back to the schedules from 1956, you can easily produce an average speed of 50 MPH on the train. That's not perfectly drive time competitive, true, but it's not horridly slow like you get in a lot of places and it isn't like I-95 in South Florida is a bastion of the open road. However, once the improvements south of Cocoa are done, I'd be surprised if you couldn't beat those schedules by a significant amount. Again falling back on the 1956 timetable and using Cocoa-Rockledge station as a convenient stand-in for the Cocoa stop for AAF (not in the least because the station is likely already in use), we find ourselves at mile marker 174 south of Jacksonville. For handy reference, Miami's station is at mile marker 366 and I presume that is a fair guess as to the location of the new station as well; that gives 192 miles south of Cocoa. West Palm Beach is at mile marker 299 (67 miles north of Miami) and Fort Lauderdale is at mile marker 341 (25 miles north of Miami). Also as a handy reference, Orlando International Airport to Cocoa comes out to 38.2 miles; OIA to that station might be another few miles, depending on where the split would come in and where that station was.
AAF wants to run trains Miami-Orlando in three hours or thereabouts. Even assuming that the AAF trains could manage an average speed of 110 MPH for those 38 miles, you'd still need to take about 20-25 minutes off the travel time there to "back up" to Cocoa...which gives about 2:30-2:40 Miami-Cocoa. Looking back to the 1956 timetable, the Havana Special did Jacksonville-Cocoa in 3:16, the City of Miami in 2:56, the East Coast Champion in 2:47, and the Miamian in 2:53. I'm going to call the time 2:50 (the Havana Special was an FEC-only train that was timed not unlike the Twilight Shoreliner in terms of aiming to keep its endpoints at a decent hour), but also note that this includes stops at Ormond, Daytona, and New Smyrna (with a ten-minute hold at the latter in most cases) that would likely be able to be dropped to make up for track conditions.
On the assumption that FEC could improve its northern tracks to match those times north of Cocoa, a through train should be able to do the run in about 5:30. That is very competitive with the interstates (especially amid traffic roulette all along I-95) and probably makes more sense than a train terminating at Cocoa.