I have read about this in on-line articles over the past two years. The general agreement is that quantities of millions of cars would lower all production costs to be several thousand more than what we currently pay for our gasoline powered cars.
Keep in mind that advances in fuel cells are coming yearly. That what a company would make today is a generation behind what it would make next year and two generations behind what the company would make in two years. I believe that with current advances, in five years, fuel cell production will be quantum leaps over how a fuel cell would be made today. Look at the advances in gasoline powered cars over the last century.
The infrastructure is bought up repeatedly, but I think it is not relevant. If the following two hurdles can be overcome, (1) an efficient fuel cell manufactured cheaply and (2) hydrogen produced more efficiently (two big ifs), the infrastructure will pop into being overnight because of capitalism.
Not only that, but there are fuel cells out there that use gasoline or natural gas. Some strip the hydrogen with a reformer, which adds cost, others don't. NASA has led the development of many of the latter type (solid oxide - see, not all of NASA is a waste of money).
That solves the infrastructure problem right there.
But I agree - infrastructure will appear rapidly IF the vehicles are market-ready, especially if the gov't helps make it a priority.
I'm not as optimistic that it will be just 5 years, but I'd say in about 15 we might see some niche applications....