I've run substantial systems (200+ node operations) on OVH and it was completely unreliable. The hardware kept breaking down, the internal network was terrible,

I've run substantial systems (200+ node operations) on OVH and it was completely unreliable. The hardware kept breaking down, the internal network was terrible, and the support was terrible. Sometimes some backhoe would cut the network connection and cause a 12 hour outage (ok, over 3 years only twice). It was also was inelastic, and provisioning was slow. So if I needed new hardware there could be considerable lead time required. We also tried Heztner, Linode and others. They were all bad in their own unique ways. With AWS, and GCP, yes it is more expensive and also have their own set of problems but nowhere near the same scale as these other providers. Yes if you are running a 1-2 machine setup, who cares. If you are running a professional organisation where you cannot reasonably have downtime without customers screaming, don't go for the cheap stuff.