Django has basically the same architecture as Rails, so it's really in the same boat from that point of view.

Not really the same architecture. It has it own differences and has much less magic involved. The project direction and the community is determined more by the community and not by a BDFL.

And equally bad at concurrency.

I don't buy this, having written and dealt with a lot of concurrent Django apps - I think people read something about the GIL often, and don't understand what the articles are really meaning.

Basically Django is usually fronted with mod_wsgi and you use celery for backend tasks. The multiple requests part are dealt with because the web server pre-forks, and it's just fine with it.

I think Django's problem is it moves very slowly, and mostly just that. The ORM isn't awesome, but it's workable. You can also choose to not use parts of it, like templates, and that's common in single-page apps.

It's speeding up development lately, 1.8 came out rather fast but it seems like the Django team likes to have ~1 year development cycle.

Exactly, the development pace has been amazing and it is going in the right direction.

It is not very difficult to plug in gevent/tornado/twisted with Django and deal with it in a pretty effective way. Also, most the resource memory intensive tasks are delegated to celery anyways.