I do want to point out that I wrote this in 2019. It was a vastly different landscape, and AWS did a terrible job of promoting the ML value proposition.
I would love to see a retrospective on the AWS uncomfortable truths as a blog post and how they've held out over the past couple years.
2019 was a different time - though I suspect that your statement about making money (as in profit) rather than just revenue (reselling compute for less than you bought it) would hold true for most companies.
2019 was a different time - though I suspect that your statement about making money (as in profit) rather than just revenue (reselling compute for less than you bought it) would hold true for most companies.