That's a sales line. Or do you work exclusively for small companies that are selling high demand online services that're trying really hard to go viral?

The only other time I can imagine is if a company isn't going to exist in six months, such as movie productions, or companies that have so little going on that they don't need more than a $10 a month plan.

Otherwise, on premises or colocation are invariably cheaper when there's any kind of real usage.

A lot of small companies have very stable workloads for years. Unless you meant startups?

> having to provision for peak load is not cost effective.

That is true, if peak load is much much higher than mean load.

For most companies, that's not the case.

On the cloud you are overpaying for cpu, network and storage by a very wide margin. That makes sense if your peak load is, say, 100x your mean load (and you have sufficienly smart autoscaling to optimize for this).

But for most companies there is not such huge multiplier from mean usage to peak usage. So those are better off with owning hardware.

It just depends on the business, b2b can be very consistent. Some workloads just have predictable behavior.

I worked at a medium sized place where we had plans in place to handle the workload for our entire market on-prem (if we were ever to capture that market).

We realized that horizontal scalability wasn't important to us because we had enough of the market to know that the other half of the market could be supported on our existing solution.

It becomes cheaper because they will run long after support has ended and cobble together a working server from the pile of broken ones, all the while rejecting getting a new server that could handle and replace dozens.