That's not the hydrogen's fundamental fault as a medium. It's the fault of the conversion process. 33 kWh of thermal to 15 kWh of work, according to your calculation. That's in the realm of highly efficient (Diesel) internal combustion engines.
As for 4 miles per kWh for a Tesla, you're already comparing apples to oranges. The Tesla feeds off batteries, where the expensive conversion part is already past. You're comparing high-exergy (battery) to low-exergy (raw hydrogen).
I don't know anything about fuel cells, but imagine they reach 90% efficiency (is that even theoretically possible? Don't know); suddenly you get 30 kWh of work out of 1kg of hydrogen. We're just starting out, and a hundred years ago ICE efficiencies were atrocious, too.
Are we just starting out with fuel cells? They were on the Apollo missions and the Space Shuttle, and Wikipedia says they were invented in 1838 and commercialised in 1932.
And hydrogen is produced via electricity, so I’m not skewing the discussion at all.
Round trip efficiency of battery is 90-95% (can be higher). Round trip for hydrogen (ie electrolysis & then storage and then fuel cell conversion back to electricity), including cooling (required for fast refueling) and compression, is about 25-40%. Even when that hydrogen is already in your tank, you only get about 60% of the energy out of it in useful energy. Heck, in California, hydrogen stations typically are only 33% renewable with most of it being gray hydrogen (9.3kg of CO2 per kg of gray hydrogen, not counting liquefaction), which makes hydrogen cars even at $10-14/kg, worse for the environment than even just random electric cars (although many EVs use primarily nuclear or renewables).
And while electric cars can have distribution losses, they’re pretty low, ie single-digit percent. But hydrogen stations are almost always refueled by truck, which has its own inefficiencies, and the stations also must constantly maintain the hydrogen at cryogenic temperatures to be ready to fuel. Also, some new stations get fuel delivered as liquid hydrogen, which is denser than compressed and therefore easier to transport, but liquefying 1kg of hydrogen (and converting it to para-hydrogen, which is normal) requires about 10-13kWh of electricity… which could send a Model 3 about 40-50 miles. So you almost can get further on an electric car on just the energy needed to distribute the kilogram of hydrogen as the hydrogen car can get by consuming that kilogram!
It’s pretty bad. Hydrogen for transport just doesn’t make much sense except for niche use cases. Maybe the last 10% of air transport that proves electrification-resistant or certain long distance ships or rockets.
The 33kWh per kilo of hydrogen - that's the absolute physical limit of green hydrogen production via electrolysis, assuming 100% efficient production, no leaks, no labor, no maintenance, no electricity losses, frictionless transportation and infrastructure that just magically pops up where needed. It's not the case that technology will push the price down - this is the best anyone can ever hope for
That's just a physical material property of hydrogen itself. It has nothing at all to do with anything else. 1kg of hydrogen has a (lower, not upper) calorific value of about 33 kWh, period. Doesn't matter where it's going, how it was made, what infrastructure was used, what any of the efficiencies involved were, ... You could magically conjure up 1kg of hydrogen from outer space and it would have that property too.