The question is not whether you pay a massive efficiency penalty for storing electricity as hydrogen (obviously, yes) but whether there are applications where you are effectively buying a set of useful properties for the losses you are "spending".
For displacing existing grey hydrogen from the chemical industry, a possible reducing agent in steel making, and other high temperature industrial processes it's pretty clear that the answer will be yes.
I happen to think that it is really unlikely that the answer for cars and vans will ever be yes, even if hydrogen could be made cheaply.
Whether it will have a role in other applications depends on a few things:
First, the structure of the future electricity mix. In order to ensure that there is enough electricity at every moment, we have to do a mix of things:
-"over" build variable output renewables so that the troughs in production are higher
-Intra-day demand side response
-Intra-day storage (almost certainly batteries)
-Dispatchable renewables like hydro
-Dispatchable sort-of renewables based on biomass
-Dispatchable fossil fuels (with CCS)
-Load-following nuclear
The reason why people think about hydrogen and other exotic technologies is that many of the technologies on that list either have limited available scope (hydro, biomass), are probably not really carbon-neutral (biomass, CCS), rely on unproven tech at scale (CCS, new nuclear designs), or have exceptionally high capital costs (any new nuclear but especially when at low load factors, CCS especially at low load factors).
So what most simulations show you're left with as a low-cost mix for most days is a combination of the first five. That still leaves you with two problems:
-What do you do if the average output net of load is very different between seasons?
-How do you provision for a 1:20 yr weather system without making your system vastly more expensive?
and one opportunity (which is also a challenge):
-variable renewables are cheap and getting cheaper so cost-optimal generation mixes build a lot more than needed for the average day in order to keep use of expensive biomass and limited hydro capacity to a minimum. As a result, electricity has low/no marginal value a lot of the time.
Second, the future cost of electrolysers. At current capex levels, they need to be run at max load all the time and still produce expensive hydrogen. Genuinely low costs like what BNEF thinks might be possible ($100/kW) allow you to run for only 20% or less of the hours in the year and scoop all the excess electricity up.
Third, the cost of alternative solutions in the areas where hydrogen is currently being considered as this reduces the value of any hydrogen produced.
I do agree that for passenger cars it's just not going to happen.