I think the problem is the focus on data science. Companies are realizing that their money is better spent on software engineers rather than data scientists, especially now that data science has effectively been "democratized".
Someone recently pointed out to me that overfertilizing young trees makes them spindly and prone to leaning or even breaking. The wood is just not strong enough.
In the 90's they scoured every math and science field for people to throw into 'web developer' positions. When the bubble collapsed a few of them had learned a lot and stayed around, the rest had to wander off and find something else to do, often back to what they were doing before, if they still could.
The data sciences growth was not organic, which means it probably isn't sustainable either.
