The problem with development of talent is that the outcomes can be so variable, and the cost so high when it fails. Identifying the individuals that are worth investing in seems to be the most important skill here, but it's extremely hard to do reliably. Nothing in my career has ever burned worse than spending years of time and effort to develop someone who simply never gets it or gives up.
Developing talent in house makes no economic sense if you can find affordable talent already developed somewhere else. Especially since your in house developed talent can bail once they got some talent.
I think the bigger point is that in a functioning economy, companies would not have any choice but to train in house, because skilled labour looking for work would not be found in excess.
I don’t think there’s any real cost to failure in this case - the worst case scenario is a high school student does a bit more extracurricular mathematics before going into a different field, where they will likely still benefit from that extra mathematical training.
The cost being the time and effort of a senior dev to mentor, which ostensibly should pay itself back as an investment over years. Sadly this rarely happens.
