The most common way to express efficiency of cooling is the "Coefficient of Performance". If COP were 100%, then every unit of energy expended (like a watt) would result in the same amount of heat removal. A true 100% cannot be achieved in the physical world in which we live... still, the closer the better.
- Compressor/refrigerant based systems have demonstrated COPs as high as 60%. A more typical real world value of an economically viable implementation is 45%.
- Peltier (solid state) have demonstrated 5%. A more typical real world deployed number is 2%.
So, yeah, there is a reason. I don't know the exact capacity of a Tesla A/C; many sedan sized vehicles are in the 24000 to 36000 BTU per hour area, max capacity. Of course, they don't run all the time; the peak is used to quickly cool a hot car and after that, they cycle, and they cycle a lot. Let's assume (pulling this out of thin air) they run 20% of the time. Assuming the same for a Tesla:
- 7 to 10kW * .2 (for cycling) / .45 (for efficiency) means a compressor based A/C will take about 3 to 4 kW from the pack spread over one hour, which makes it very easy to say 3 to 4 kWh. That's about 4 to 5 % of the total pack (i.e. reduction in range, per hour).
- If Tesla invested in a very efficient compressor based system, like 55% efficient, that could be as low as 2 or 3% of the pack.
- 7 to 10kW * .2 (for cycling) / .02 (for efficiency) means a solid state based A/C will take about 70 to 100 kW from the pack spread over one hour. That's the entire pack.
Those numbers are based on a lot of assumptions and averages. Reality in a Tesla may be different... but the relative efficiency will be somewhere in that ballpark, namely "a few percent" vs. "pretty much everything".