When it comes to converting one form of energy one way to mechanical work only a Pelton wheel can approach an electric motor's 95+% efficiency.
In sharp contrast all single cycle high compression ratio ICE small enough to be hauled down a road tops off at 40% efficiency with vehicle size spark ignition ICE generally running below 30%. That's a well tuned engine running on it's "sweet spot" rpm.
Why can't anyone do much about the sorry efficiency of 99.9% of the prime movers on the planet?
Far and away the biggest problem comes from the basic thermocycles of adiabatic engines, i. e., gas turbines (Brayton/Joule/Ericsson I), diesel (Diesel) and spark ignition (Otto).
The machinery dictates the processes and the shape of the thermocycle and even the idealized [read: fantasy] adiabatic cycle doesn't fill a Carnot or other isothermalized parallelogram cycle -- the thermodynamic limit of heat engines -- very well. Materials temperature limits reduce the Carnot limit below 100%, so maybe a little over 50% Carnot for most ideal adiabatic cycles.
The real cycle, however, looks more like a paramecium. The nice sharply defined corners of the ideal have been rounded reducing efficiency still more.
Toss in incomplete combustion and other parasitical losses and electric motors start to look pretty.
Bret Cahill