I have no clue about the reliability of OVH, but I would definitely not say AWS is short on failures. At a scale of ~2000 instances, it's really not uncommon to encounter a bunch of AWS instance failures every week. Multiple times per year we see a massive failure that affects a large number of servers. The saving grace with AWS is that usually the problems are limited to a single availability zone so as long as you're redundant across multiple availability zones you can survive.