> they offered a handful of high-value services, all at great prices, making them cost-competitive with bare metal but much easier
That was never the case for AWS, the point was never "We're cheap" but "We let you scale faster for a premium".
I first came across cloud services around 2010-2011 I think, when the company I worked at at the time started growing and we needed something better than shared hosting. AWS was brought up as a "fresh but expensive" alternative, and the CTO managed to convince the management that we needed AWS even if it was expensive, because it'll be a lot easier to tear up/down servers as we need it. Bandwidth costs I think was the most expensive part of the package, at least back then.
When I look at what performance per $ you get with AWS et al today, it looks the same, incredibly expensive for the performance you (don't) get. Better off with dedicated instances unless you team is lacking the basic skills of server management, or until the company really grown so it keeps being difficult dealing with the infrastructure, then hire a dedicated person and let them make the calls for what's next.
I'd agree that AWS never sold on being cheaper, but there is one particular way AWS could be cheaper and that is their approach to billing-by-the-unit with no fixed costs or minimum charges.
Being able to start small from a $1/mth bill without any fixed cost overheads is incredibly powerful for small startups.
If I wanted to store bytes in a DC it would cost $10k/mth by the time I was paying colo/ servers/ disks before I stored my first byte. Sure there wouldn't be any incremental costs for the second byte but thats a steep jump. S3 would have cost me $0.02. Being able to try technology and prove concepts at the product development stage is very powerful and why AWS became not just a vendor but a _technology partner_ for many companies.
> Being able to start small from a $1/mth bill without any fixed cost overheads is incredibly powerful for small startups.
Yes, no doubt about it. Initially AWS was mostly sold as "You never know when you might want to scale fast, imagine being featured in a newspaper and your servers can't handle the load, you need cloud for that!" to growing startups, and in that context it kind of makes sense, pay extra but at least be online.
But initially when you're small, or later when you're big and establish, other things make more sense. But yes, I agree that if you need to aggressively be able to scale up or down, cloud resources make sense to use for that, in addition to your base infrastructure.
But if AWS didn't have that anti-competitive data transfer fee that gets waived if your traffic goes to an internal server, why would you choose S3 vs a white-label storage vendor's similar offering?
> the point was never "We're cheap" but "We let you scale faster for a premium"
Actually, it was more like "Scale faster, easier, more reliably, with proven hardware and software infrastructure, operated by a proven organization, at a price point that is competitive with the investment you'd have to make to get comparable hardware, software, and organizational infrastructure." But that was then. Today, things are different. Cloud services have become giant hairballs of complexity, with plenty of shoot-yourself-in-the-foot-by-default traps, at prices that can quickly spiral out of control if you're not on top of them.
Yep.
AWS is a money printer because it enables companies to make a lot more money without hiring multiple 200k a year engineers to manage infrastructure that comes with scale. And that is also considering that the company will actually have to find competent people to do the work.
With AWS, if you wanna scale, you just use your credit card without worrying about anything.