There's a great essay[0] by Steve Yegge that talks about a certain spectrum of philosophy amongst developers with those liking lots of compile-time validation at one end and the free-and-easy dynamic runtime-checked-if-at-all ones at the other.
Those of us at the "conservative" end of this spectrum see great alignment between what the rust language offers, what the rust language development process promises, and what we're seeking. Seeing that, there's also the tantalizing prospect of like-minded peers amongst the enthusiasts for it.
It's also just kind of neat from a technical point of view that it's possible to make the sorts of rigorous promises at compile time that rust offers.
Understanding the borrow-checker rules is a famously tricky challenge for newcomers to the language, which perhaps nerd-snipes[1] some of us.
Because it's a systems programming language (writing an OS directly in Rust is perfectly possible [2]) it has some allure as a "grown up" language.
The rust language development community is famously welcoming with good quality documentation and the IDE support is getting there. So it's approachable. That helps for sure.
Finally there's unarguably some evangelical fanboy-ism going on that no doubt boosts the numbers a bit on top of all that.
Disclaimer: I'm not a rust developer, but I'm definitely attracted by what it has to offer :)
---
[0] Alas it was posted on Google+ so the original is no longer up, but there are a variety of mirrors of it out there, e.g. https://gist.github.com/cornchz/3313150 - see also meta-discussion here: https://news.ycombinator.com/item?id=4365255