Prices should have gone down because they've continued to scale up and go into custom cooling/hardware/backbones/etc.

You're basically saying that you're cutting them slack because their margins have gotten much larger and they haven't decided to raise prices to make them even bigger.

Why should that make prices go down? Prices aren't based on cost. They're based on what the market will bear, i.e., what customers will pay. If you think AWS is too costly, you can go find some other solution. If enough customers start doing that that AWS's financial state suffers, AWS will have to rethink its price and cost structure.

Competition in general tends to drive prices down, yes; but competition was not what the post I was responding to claimed "should" make prices go down. That post was claiming prices should go down simply because AWS is investing more resources in building capital as they scale up. That is not true.

> The fact their prices and profits have remained high for so long indicates lower competition, likely due to high barriers to entry from competitors and high switching costs for consumers.

It could, but it could also indicate that this market simply has not reached equilibrium yet. I don't think it's plausible to claim that AWS has no competition in this area, or that barriers to entry are high; there are a number of huge corporations investing a lot in this market, and plenty of smaller players gaining customers by presenting a simpler interface to an AWS back end (e.g., Heroku). Switching costs might be high, not just for AWS but for any provider in this market, simply because there is so little standardization in how infrastructure is specified and controlled. That would lengthen the time to reach a competitive equilibrium in this market.

It is worth remembering that for new technology markets, it can take a long time for a competitive equilibrium to be established. A good example is the automobile market; in the US, for example, it took many decades for car prices to be driven down to marginal cost and for all of the various market players to search out and capitalize on all the possible competitive efficiencies and strategies. Part of that was also that it took decades for the auto market in the US to become saturated, i.e., for most new car sales to be replacing old cars instead of getting a car to someone who had never owned a car before. I don't think the market for AWS-like services is anywhere close to saturated, which means we should not expect a competitive equilibrium; instead, we should expect exactly what we see, large market players trying to capture as much market share as possible during the growth period, just as major automakers did in the US during the growth period in the mid-20th century. Capturing market share in a growth period is a very different game from squeezing out efficiencies in a market near competitive equilibrium.

My point is that their prices could have gone down without them even feeling any pain because their costs have gone down. They don't deserve slack for keeping prices the same. Keeping prices the same just means their competitors either suck or they sit in an oligopoly together.

The slack I'm giving is for the complexity in their billing, not in the price itself.

You always need the head to subsidize the tail (e.g. free trials). You can’t let the head become unprofitable.

And they can also impose things like bandwidth caps to curb some less-profitable customers' usage. But cloud providers, by design, are charging "per use" and expected to scale "indefinitely"... at least for expected use cases.