Since violation of conservation of energy is very unlikely, I'd say that this is a sign that the drive doesn't actually work. Or at least that there is some gotcha that we haven't understood yet.
The EM Drive is an energy reaction engine, rather than a mass reaction engine (if it works as described), so instead of pushing off wake, it uses energy itself to "deepen the fold", as it were, and increase speed (and mass, since the two are exchanged).
That's interesting. I've never considered that the very act of increasing the velocity of the EM Drive would actually increase its mass. Hrm.
And anyway, kinetic energy isn't a conserved quantity. It can be converted into potential energy in a gravity well, or lost entirely in inelastic collisions. Momentum is the quantity that is always conserved, and that is what the emdrive violates directly (though still not in a way that allows for perpetual motion or free power, near as I can tell).
The first internal combustion engines were quite inefficient as well.
The advantage of the emdrive is that it does not require reaction mass, at the cost of incredibly high power consumption. Unfortunately that makes it not really viable for anything; ion drives are limited by power, not reaction mass, so the emdrive is actually worse because it uses way more in power than it saves in not needing reaction mass.