It cost a HUGE amount of money. The reason is that in the plane you have people, so the rocket design has to take that into account and validating the rocket costs more. Just for reference, SpaceX and RocketLab sent their rocket to space for under 100M, while Virgin spend like 500M. That's partly because of worse leadership, but its partly inherent.
A plane also isn't cheap, a ground launch station for a small rocket isn't that expensive.
Additionally engineering the rocket to get carried vertically adds structure, that cost money and lowers payload. You end up not gaining that much from air-launch. It also makes the structure and avionics quite complex.
Once you drop the rocket, the engine has to start or the payload is gone. Other rockets can start the engine, verify its good and then launch. Relativity space just did that.
It also puts huge constraints on your rockets, the size in both weight and volume is inherently limited, and you can't evolve the architecture. In comparison SpaceX Falcon 1 could have easily been slightly upgraded and turned into a rocket that could lift 1T+ and even more eventually. And then they managed to use that engine to build a Falcon 9.
Virgin Orbit has this one rocket, and its impossible to make back 100s of millions of investment with a tiny rocket and low launch cadence. They can't make the rocket significantly bigger because of the limits of the plane.
I thought it was impossible for them to ever be a real company and that seem to have ended up happening. Launch is a really tough business.
> And why does a single failure doom a company after 4 successful launches.
Because they have already burned down a gigantic amount of cash and were in a bad stage, launch success or not. A launch earns them 12M or so. That basically nothing considering their investments and their burn.