All of those are true. The lack of powerpoint at Amazon is not a myth - docs are the order of the day. This is more good than bad, but has some of both.
EC2 has been in operation for a few years now, and the price levels from the launch remain unchanged, despite the fact that their largest expenses are subject to Moore's law.
The price levels of everything except for running an instance for an hour have dropped significantly since launch. This includes bandwidth (the pricing tiers used to be $0.18/$0.16/$0.13 per GB and have since fallen to $0.17/$0.13/$0.11 with the addition of a fourth tier at $0.10/GB), S3 storage & requests (used to be $0.15/GB for storage and $0.20/GB for requests, now is $0.12 - $0.15 for storage and $0.10 - $0.17 for requests), Simple Queue Service (the changes here are more complex but Amazon writes "Under the new plan, 76% of customers with bills greater than $1 would have received lower bills, saving an average of 71% each compared to their actual bill", so it's generally much cheaper), Simple DB storage (used to be $1.50 per GB-month, now is only $0.25 per GB-month), even the Cloudfront CDN has already gotten cheaper since its very recent launch date (the lowest priced tier used to be $0.09 per GB and they've since added tiers from 5 to 8 cents per GB).
So while you're right that the instance-hour cost has remained flat since the launch, the actual costs of using the AWS platform as a supercomputer or server farm has fallen dramatically.
Well, putting aside the fact that you are pretty much wrong (servers are cheap, power, cooling, etc, not so much), keep in mind too that their initial pricing may not have been profitable, or may even have been at a loss because what matters is the total lifetime value of a customer.
Amazon (or any other hosting provider) may set their initial price below cost because declining costs will give them a reasonable margin when figured over the space of a few years.
Even mathematically inclined people get brainbusted when someone says the I-word, but I have to try: if you are able to buy some commodity for $1 in 2005, and you are able to buy the same commodity for $1 in 2008, then the real price has declined by about eight to ten percent, give or take.
The point of the cloud is that it is automated and the development costs are negligible after being amortized over so many servers. If EC2's largest cost is salary then either they haven't reached sufficient scale yet or they're somehow doing it wrong.
I enjoyed the story thanks for sharing, a couple things I didnt get is why the author "was trying to avoid this.." and why "after some exchanges its best to share"
- I bet he feels some ownership, over the product.
- The 2nd option is this is a resume` play, but probably not likely.
either way great story; although short - good luck with your other endeavors
After a project succeeds in a large company, people who were tangentially involved come out of the woodwork to claim credit. There is a natural human urge to debunk these claims, but often this just makes things worse if the credit-takers are politically astute.
1) This came up b/c somebody else publicly described the author as the inventor of EC2, which created some unintended blowback;
2) The author is well-known in the ops community and still close to a number of people at Amazon, and has no personal or professional interest in hurting those relationships.
I do think they should have picked a different brand name than Amazon for it.
We're moving to EC2 at the moment and when explaining it to commercial people it's an unnecessary hurdle to also have to explain that Amazon do more than sell books and they really can be trusted to host infrastructure.
1) "we presented the paper to bezos" -- Wow, engineers have direct access to the CEO.
2) "we presented the paper to bezos (he doesn't do slides)" -- Wow, a CEO that actually has some attention span
3) "we presented the paper to bezos (he doesn't do slides), he liked a lot of it" -- Wow, a CEO who actually understands technology