Before the cloud, there were many bugs in custom-built software that could be have been solved with more memory or more CPU. But as getting more memory and CPU was incredibly expensive (buy new servers, receive them, rack them, connect them, install all the stuff in them) it was rarely a real solution. Devs had to find a solution to the constraint.
Now we have the other way around, adding memory or CPU is a few clicks away, so solving a software issue by adding more CPU/RAM is easy.
How much lousy code is cloud helping to hide?
One day the harmful code will show its true self and crash a bigger server, destroying more data and impacting more users. But until then costlier hourly rates will allow it to be undetected.
Make sure you have a culture that cherishes good quality work, recognise such quality and does not blame when things go south (they always will). Invest in quality people, great developers (or train them to be), adequate quality assurance.