This may be slightly different from a political cycle that spends ever-increasing amounts of resources on roads without examining if the results obtained provide a reasonable level of value for money. Or indeed address the issues that drove the construction of additional roads in the first place.
Broadly, fully utilized resources are not always an ideal goal. For example, my experience is that meeting rooms in offices are almost always in constant use, no matter how many there are. This may not be the same as all uses of meeting rooms being desirable and efficient uses of the time and other resources used. I know I have been in meetings, enabled by a sizable of supply of meeting rooms, that could have been handled much faster with a handful of emails.
Have you ever worked on a shared compute cluster? In general, disks fill to capacity & machines run at capacity, no matter what those capacities are. It's bloat; files that don't get deleted until you run out of disk, because deleting files takes ten seconds. Massive processes that are left always-running to save ten seconds of launch time. So on and so forth. Basically, the user is getting some value. But it's very small, and thus makes a poor reason to make expensive upgrades to the cluster.
Saying "traffic will be solved by this!" is clearly misleading. But those extra cars on the road must be benefitting, or else they wouldn't make the trip.