Node processes its own logic in 1 thread, but it still has multi-process support. So you can use clusters to have multiple instances running at once, or go old school with spawn/fork. It does confirm what you said: it is an old technology as far as multi-core goes, and you're not going to be writing a ray tracer this way.
For I/O intensive load though, just use clustering to have 1 instance per logical core and you're golden.
Node's claim to fame is really just its async-first I/O design. Some other environments are bringing that in now, along with a strong threading model (ie: .NET, between almost everything new being async by default, and the async/await coroutine syntax), but the culture/ecosystem around it aren't following. Also, if your only easily parallel tasks are I/O bound, you're not getting much more out of it than you would in node.js.
Node however, is NOT simple, it just looks simple. Its absurd stream design and error handling alone means that on any reasonably complex application, you need to be one hell of a debugging guru to figure out whats wrong. A junior dev will run home crying, while the same things in most other languages are pretty trivial.
If you're not building a scripting tool (ie: a css compiler or something), don't need javascript on your server for isomorphic apps and you don't benefit from the threading and I/O model in your workflow (ie: your app is CPU bound), stay far, far away.