You're only really paying for the cost of the hardware, the rest is a heavy reliance on peering with no or little transit links (you'll see they only sell to some EU countries).
On the forums the only person complaining where people doing NSFW hosting or radio streams. For web app stuff (including Git/Subversion hosting) it is really really good.
You can check the backbone here: http://weathermap.ovh.net/
Also, for people not aware, they have their own servers with liquid cooling and they are crazy with respect to optimization of their processes, think TPM and Toyota.
The concern is, just because it works for you, doesn't mean it will work for all of your customers. Especially if they're not in the EU.
I'm not so sure on the cooling either -- their hard disks are very hot (I admit I don't know enough about hardware to say whether this is a problem) but generally I don't see many hardware failures.
I do however see plenty of random reboots -- only the HG servers (and strangely a few 19.99 servers with over a year uptime) seem to survive these as you'd expect.
I've taken new servers with Leaseweb and I'm building up with them now, I also have a few with 100TB in the US too. So now that when a customer says they are having poor speeds I just need to verify that it's to do with the bandwidth and explain the situation and this works very well for me. I can then move them to a new server and everyone is happy, it's an easier way to deal with something other than "err, sorry, would you like a refund?"
Although I realise I'm unique in that I can just move them around (well, 95% of the time). What would concern me is running a business critical website on a provider that doesn't provide international bandwidth.