There's a lot of truth in this. Particularly since the free tiers on most cloud providers are explicitly designed to suck you in, then raise the rates tremendously as you scale.

A business that might be fine with 100 customers on the cloud could struggle mightily to cover the costs of dealing with 1000 customers on the same stack. The linear revenue growth doesn't match up to the insane cost increase as you go from "hobby project" to "small business" in the cloud.

Sure - you were hosting your DB just fine with 100 customers using standard SSD, but now at 1000 you're blowing your IOPS budget and need to upgrade to premium SSD. The premium SSD costs twice what the standard SSD does for the same space - your usage will not have doubled.

Or - god help you, you were hosting your db just fine with 100 customers using a small VM, and to avoid having to think about it, you've decided to move from a self-managed db on a VM to something like Cosmos DB. No growth in usage at all from you, but your costs are about to shoot way up.

Or - I've added a customer in a region that has poor connectivity to my current region of choice - oops, setting up edge services closer to them is going to hurt. Now I'm sending data between regions, incurring all sorts of costs between services that I wasn't with a single region setup.

----

My experience with the cloud (and I've been a heavy user of both Azure and AWS, not as much GCP) is this: There are bands of costs, within that band, costs will tend to grow linearly as you grow.

The problem, is that you will outgrow the band you're in. There is a hard limit on something that you weren't thinking about (IOPS, for example) and once you hit that you have to jump to a different band with a new cost structure.

Usually, that means increasing base spend by a large amount compared to usage, and then getting slightly better cost/use at the new higher base.

Everything dies. This simple minded confidence game we require each other play to validate usefulness is getting old.

You’re one of seven billion humans who will die, one of millions of programmers to pick from to prop up trade of Daddy Dollars.

You’re not sticking around either and cannot prove your contributions were truly more valuable than anyone else’s (models have been built to ascertain which human leadership traits and which technology make us more productive and all became too complex to understand).

Don’t blink though; keep pretending we’re onto something with folksy anecdotes, and nostalgia for past wins.

Sure, it depends on what you are doing, but I imagine that for most paid SAAS models, the cost of cloud computing is a small fraction of what they bring in revenue. In many cases, so small as to be effectively $0. As an example, stack overflow gets hundreds of millions of visits a month, but is run on two physical servers (plus an additional two for backup). As a more personal example, I have a small SAAS side project and my server costs run me about 0.5% of my revenue (the rest is all profit other than the value of my time). I could probably get that down to 0.1% with some code refactoring, but even with the current code, I could easily 10x my number of subscribers and not increase server costs by even $0.01.

To add to this, in my experience, the companies I've worked for have tons of extra capacity with their cloud services. For example, using t3.large vs t3.small instances at 4x the price indiscriminately. Why? Because server costs are just a rounding error compared to paying a bunch of $100k+ salaries.

Regarding it costing money, yes, of course. But costs increases as users increase. If you're giving away your product for free, that's an issue. If you're charging, then I don't see the problem.

AWS has a free tier, so I wouldn't apply that rule too much.