Hands down, my vote is for Palantir Technologies. It's underhyped because it's not a consumer or web play, but the strength of their engineering team and the scope of the problem they are trying to solve makes it a very interesting company.
It's not just about data visualization; it's attacking the problem of how we navigate, learn from, and make inferences from large sets of data. The interesting thing is that while Palantir uses a lot of ideas from machine learning, its fundamentally a tool that offloads the task of "AI" to the best machine for that function – the human brain. It's a hard problem, and not just one of UX or interface design. What it requires is a new way of choosing what information to surface, why to surface that information, and what inferences one can make with that data to further exploration of the dataset.
And that's a separate problem from the (extraordinarily difficult in itself) challenge of efficiently storing, operating on, and cleaning huge, rapidly-changing datasets.
Palantir has developed technology that has significant 'real' option-value apart from government contracting.
Palantir is awesome 🙂