The fixed cost problem with the hub and spoke model is mainly due to the need for banked flights to minimize average connection times.
Ideally, every plane starts the morning at a spoke and arrives at the hub at roughly the same time. Then passengers have an hour or two to change planes, and all the planes fly back to their spokes. This cycle is repeated again in the afternoon, and then the planes spend the night at their spoke to start the cycle again the next morning.
The problem is that this requires duplicate gate space at the hub and spokes, even though planes will only use one or the other at any given time (or neither, while flying). It also requires sufficient planes to serve every spoke in every bank even if there’s not enough demand to fill them. And it requires much larger airports and terminals at the hubs to serve massive numbers of passengers and flights during the banks, which goes unused the rest of the day.
Of course, nobody operates quite this stupidly in practice, but every flight that deviates from the model means longer connection times for its passengers because if it arrives at the hub between banks, passengers have to sit in the terminal and wait for the next bank to connect, which makes your flights less desirable than a competitor’s.
The upside is that you can (in theory) size the aircraft serving each spoke to its average demand, and likely use fewer, larger aircraft since one plane can serve every passenger connecting to every destination, which reduces variable costs.