This boils down to well-known and well-understood behaviours of energy systems, particularly at large scale.
Much of the waste is simply Carnot efficiency of thermal systems: the amount of useful energy you can extract is proportional to the (absolute) temperatures (Kelvin) of the "hot" and "cold" ends of the cycle.
For thermal electrical generation -- coal-fired plants, gas-fired turbines, oil-fired turbines or diesel generators, biomass thermal, solar thermal, or nuclear plants -- this is in the range of 30 - 45% or so. Ironically, high-temperature thermal coal achieves some of the best thermal efficiency. This doesn't mitigate its far more compelling downsides.
Direct kinetic or photovoltaic electrical generation has no thermal losses, but is subject to the efficiency constraints of the input stream: hydroelectric, wind, or solar PV.
There are additional losses in transmission (about 6%), and in electrical conversion and switching equipment ( ~<10%). The net is about a 66% energy loss in what's delivered to the electrical customer.
For transportation, you have the same Carnot efficiency limits, but given the smaller temperature differential start with a lower initial efficiency -- about 30%. There are additional losses through parasitic systems (any powered in-car features: power steering, brakes, A/C, electrical and electronics, etc.), transmission losses, tire and wind drag.
Again, all well understood and modeled, and well-behaved in large-number populations.
The Lawrence Livermore National Lab (LLNL) has done energy modelling for the United States since the 1970s, and has a set of flow diagrams (Sankey diagrams) showing flows dating to the 1950s, through the present.