At the federal level, Congress has been transferring money from general funds into the Highway Trust Fund every year since 2001, because the fuel tax has gone down in real terms every year since 1993.
Tolls are nice but only account for 5% of road funds in the US. To use the pejorative applied against transit projects, the "fare recovery ratio" for roads ranks below literally every transit agency in the nation.
In short, roads are highly subsidized from general revenues.
But I don't have good data. I want good data but don't have it.
E.g., there are a lot of roads and bridges that are not Federal, e.g., not the Interstate highway system. So, when I hear that the gas taxes don't pay for all the roads and bridges, I think, of course not -- e.g., my local roads and bridges are paid for heavily, maybe mostly, by local real estate taxes, and in my area that is fair -- everyone here needs the local roads and bridges.
Also for the 100%, there are some toll roads and bridges where the users pay for use of the roads which looks like users paying and not a subsidy.
There's likely another issue, a law about 100 years old passed in part to slow down US West water resource projects to "make the desert bloom" -- yup, can do that, but the question was, do the benefits exceed the costs?
So, the law established that a Federally funded project needs to pass cost-benefit analysis. So, add up all the costs and all the benefits "to whomsoever they accrue", and then to go ahead with the project the benefits have to be higher than the costs.
Well, maybe passenger rail can pass cost-benefit analysis in some cases and, there, justify a subsidy, but the one case I heard about was in a lecture on the Baltimore Subway. It was all built and ready to go. So, for the cost-benefit analysis, just call the construction cost $0.00 -- seems generous enough!
Then have to count the operational costs and consider the ticket revenue. Well, the ticket revenue from the estimates of what people were willing to pay and the number of such people still didn't cover even the operational costs. The lecture ended with the optimal solution -- brick up the openings and walk away.
That seemed a bit extreme: Instead why not take out the tracks and use the rest as underground parking? Warehousing?