You have to remember the time constraint on drug profitability (~15 years for a US patent on a new molecular entity).
A new antibiotic would be held in reserve except in exceptional circumstances. Even if it was put to use on day one for extensively drug resistant TB and the like, there are too few patients to find a functional price point. It's sort of like the situation with many rare diseases.
My favored approach for solving this problem is using some sort of prize system (Sanders had a bill to do this for all drugs last year if you want to see a model). Effectively, the government offers $X billion for a new antibiotic meeting criteria Y and Z, ensuring that the private sector has financial incentive. In most plans, the drug IP would become public, allowing for immediate low cost generics.
If you develop a new antibiotic then it won't be used, it will be held in reserve for when current antibiotics no longer work. We don't want to expose bacteria to new antibiotics until we have to. This is why it would be a commercial failure.
A change to the model where the government(s) develop their own antibiotics so that they can sit on them might be in order. It's not much of a vote winner but is as essential as defence.
Develop them on a prize basis or as government IP is one way.
Another way is to shelve the start of the patent window along with the drug. If we know it works but it's not needed yet, don't start the countdown to patent expiration until it is needed.
The former perhaps lowers costs for the drug more effectively. The latter gives more incentive to the drug companies to keep doing their research.
Suppose you have a new antibiotic that effectively kills multiresistant pathogens. The correct action in this case would be to hold it back and only use it when all else fails. This means you sell at very low volumes, and even if the prices are astronomical revenue is fairly low. Now, in addition, you can sell it on the side in places with a less moral or more weakly enforced regulatory regime, and you end up with overprescription, undercompliance, and use on livestock. As a manufacturer, you can recover costs and make profit much more easily in the second scenario, so it happens whenever the pharma companies can get away with it, which is every time. If you were to force them to hold back the new antibiotic for last resort cases only, it would indeed be guaranteed to be a commercial failure, since volumes are not sufficient to justify bringing it to market.
Using a current example, Fidaxomicin, it's not currently the most widely used therapy for C. difficile infections, but it is used - and admittedly the price is very expensive. But there's also very little evidence of it being used in grey markets or for off label uses.
And I'm unconvinced its destined to be a commercial failure given Merck is pushing money into it.
Even if something is being used as a "drug of last resort" may have fairly high volume. For example, colistin, which is one of those drugs, are often used in combinations with other antimicrobials to help combat resistance, and is pretty much universally available to hospitals in the U.S.
This is not axiomatically true.