3000 rpm (at 120-130 mph) on first serves and sometimes exceeding 5000 rpm on 2nd serves could have been a factor.
http://wings.avkids.com/Tennis/Project/usspin-07.html
This is something I've talked about a lot over the years. A lot of people just go by the speedgun to determine effectiveness, but look at Henman's 120mph serve and Sampras' 120mph serve:
Player: No. of Serves: Avg MPH: RPM Range: Avg RPM:
Marcelo Rios 2 92mph 3000-3333rpm 3167rpm
Jim Courier 6 108mph 2500-4054rpm 2842rpm
Todd Martin 7 98mph 1667-3947rpm 2798rpm
Tomas Muster 8 105mph 1667-4284rpm 2754rpm
Pete Sampras 11 120mph 2100-4260rpm 2699rpm
Petr Korda 7 101mph 1579-3750rpm 2688rpm
Andre Agassi 9 102mph 1200-4284rpm 2449rpm
Mark Philippousis 3 123mph 1765-2830rpm 2198rpm
Michael Chang 7 112mph 1000-3750rpm 1677rpm
Tim Henman 2 120mph 1429-1667rpm 1548rpm
Look how much more spin Sampras puts on the ball, sometimes almost 4 times as much spin as Henman's. Also, it is conceivable that some of those could be 2nd serves because from 1998 onwards he was hitting 2nd serves in that range as he was coming to net more, pre 1998 he was staying back more on 2nd serve and didn't hit it as hard but still deep.
So basically, that's why the Sampras was serve was much harder to return than Henman's, and many other players who served faster than him during that period in tennis history. Not only did it have much more spin, but his serves were either sliding away from the opponent or swinging into the body. Sampras would have made a very interesting fast bowler in cricket because he gets so much movement, and movement translates as deception.
Just one other important thing to point out, Sampras was doing this with natural gut, not fancy luxillon strings - so technique and skill is just as important as technology in the right hands.