Doing some napkin estimating: 1000 users loading a page each per minute is 16.67 page renders per second (or 60 ms per page assuming one cpu thread). If they're reloading every 10s that's 10 ms per page. At that rate even a compiled language may not keep up with the db being the bottleneck.

That said, if any of the page renderings can be shared between users, such as using a framework that can do page fragment caching then you can approach that level especially if you can cache entire responses using nginx.