From http://www.dvr.com.au/injectors.htm :

  "Note that when the steam mixes with the water and condenses,
  this is when it forms the vacuum."

Unfortunately, the compressed air does not condense, therefore this
cannot produce vacuum.  What you would therefore need is such a high
pressure difference that the compressed air is actually _liquified_
(which is possible -- this is one technology for liquefying gases),
but this might be a little more expensive than a simple vacuum pump!

What _might_ produce vacuum is the fact that the expansion of the
compressed air after the nozzle cools down the air.  However, it is
still compressed air that you blow into the "suction chamber"; if the
whole thing should become a vacuum-creating machine, you would have to
get out the air of the chamber at the same rate as you blow it in.

The method used in a steam injector is the mass inertia of the
steam/water mixture.  To use the same on a pure gas (like air), you
would need a very carefully designed exhaust so that you get a perfect
laminar air stream out of the chamber touching the exhaust walls.
This can be done (aeroplane turbines come to mind), but the mixed-in
suction air stream will tend to create turbulences; thus, both the
inlet for the vacuum and the mixing of the two streams would also have
to be designed very carefully.

In the end, I think it could be done, but it's (a) very difficult,
(b) very expensive, and (c) it would probably have a much lower
efficiency than pumping methods (that might change with higher
volumes).

So much for a purely qualitative judgment.  If someone has really done
this, I would very much like to know whether my considerations above
turn out true or false.

Regards

H. M. Mueller
Grafing b. Muenchen
Germany