In the context of the business area that Palantir is in, though, they absolutely do need machine learning to maximize the value of their product. Imagine trying to build a program to beat AlphaGo, or a search engine better than Google, or a best-in-breed image classifier, without using machine learning. You can't do it. Machine learning is a critically important technique that can solve problems better than anything else can. Palantir's work contains many such problems. Not using the best tool for the job, and not even having a roadmap to do so, means they're ripe for disruption by some other company coming along doing the work better.

Machine learning isn't just some buzzword used to attract VC money. It really, truly, does work.