If this is just a one-time read of the whole file, and you are doing this on MS Windows with Excel installed, then xlsread() is about as fast as you can get. xlsread() does have some overhead for matters such as figuring out worksheet names, so it is possible to set up for reading a bit faster by hard-coding that kind of information, but once that is set up, the ActiveX connection works about as fast as could be.
You could also experiment with readtable(). For xls files the binary format is examined and parsed somewhat efficiently, but the code is at the MATLAB level so using ActiveX would typically be more efficient because Excel is compiled. For xlsx files when Excel is not available, readtable() uses regexp() to parse the text after having to go through a series of set-up steps, and although regexp() is one of the faster operations in MATLAB, this is still going to be slower than using ActiveX to Excel.
