Hacker News new | more | comments | ask | show | jobs | submit login
AWS Drives More Than Half of Amazon's Operating Income (lightreading.com)
411 points by petercooper 21 days ago | hide | past | web | favorite | 115 comments

“Percentage margins are not one of the things we are seeking to optimize. It’s the absolute dollar free cash flow per share that you want to maximize, and if you can do that by lowering margins, we would do that. So if you could take the free cash flow, that’s something that investors can spend. Investors can’t spend percentage margins.” “What matters always is dollar margins: the actual dollar amount. Companies are valued not on their percentage margins, but on how many dollars they actually make, and a multiple of that.” “When forced to choose between optimizing the appearance of our GAAP accounting and maximizing the present value of future cash flows, we’ll take the cash flows.” Jeff Bezos is very focused on this “absolute dollar free cash flow metric.”


People really need to get off of the "income analysis" trope with Amazon. It's not how they operate their business, so why do people (analysts and media mainly) analyze it any other way?

There's a tradeoff here. Bezos is of course a world class executive and founder, but he's underselling the advantage of big margins which is risk.

In a downturn, a slight decrease in margins across the board will push a company on the razor's edge from profitability to unprofitability. Margin means margin for error - when the company isn't doing well and investors are scattering to safety and liquidity is nowhere to be found a higher margin gives your company a buffer when you have to cut.

Of course when revenues are growing like crazy and the economy is gangbusters it sounds great to cut margins for absolute growth but that's a booming economy. Things are very different in a downturn.

Using systems theory, i think margin is a buffer for profitability that (to your point) allows for shock absorbing (as all buffers do.) Higher margins are also a risk of losing market share to a price-competitor, and you can suggest that market share is also a buffer for consumer engagement -- the ability of a firm to successfully raise prices (in the short term) is proportional to its market share.

Does the buffer of market share outweigh the buffer of margin?

Interestingly, the "what companies are valued on" is an increasingly surprising thing when you think about how real estate development works. Properties are valued on a multiple of their Net Operating Income, which creates a scenario where capital improvements that have low ROI in terms of NOI are actually high ROI when you include the effect on equity (future sale price.) In the current environment of 4% cap rates (NOI / Sale price) each additional dollar of income increases the value of the property by $26 (one dollar for the income and 25 dollars for the sale price bump.) This is why every landlord is very incentivized to borrow as much as possible to improve property as much as possible to raise rents as high as the market will reliably bear (vacancy kills your NOI and scares off acquirers) and then sell the property.

There's also an asymmetry of information and of control. In a downturn, Bezos has full visibility into where all of Amazon's COGS is going, and a number of levers he can turn to axe some of that spending, reduce future investment, or shed unprofitable business lines. Wall Street has virtually zero of that information (they get the consolidated financial statements, but none of the detailed breakdown) and zero of that leverage. Therefore, it's significantly less risky to Bezos to run on razor-thin margins than it is to an Amazon shareholder.

Same reason that founding a startup is significantly less risky for a world expert in its problem domain than for a random investor in that startup. The founder has full knowledge of the domain and a lot of levers they can pull to adjust strategy; the investor has none of that. That's also why many VCs require terms to reduce their risk (liquidation preferences, participation, anti-dilution, warrants) or increase their control (board seats, voting rights, information rights).

> when the company isn't doing well and investors are scattering to safety and liquidity is nowhere to be found a higher margin gives your company a buffer when you have to cut

And how has that worked out for 440 companies that used to be in the F500 that no longer are there?

> In a downturn, a slight decrease in margins across the board will push a company on the razor's edge from profitability to unprofitability.

Except the reason their margins are so low is because they keep feeding capital into the business to fund new projects.

> Of course when revenues are growing like crazy and the economy is gangbusters it sounds great to cut margins for absolute growth but that's a booming economy. Things are very different in a downturn.

Yup, you're right, things are very different in a downturn. Great listen about how NetSuite navigated downturns: https://a16z.com/2015/05/15/a16z-podcast-why-saas-revenue-is...

This seems like a dilemma between a superior strategy, but only if executed with discipline.

In reality the majority of businesses which run on large margins get lazy, allocate capital ineffectively, and underinvest in R&D.

Thin margins force you be lean. Thick margins afford you the opportunity (and risk) not to be.

Amazon retail has gross margins of 37% and net margin of, lets say 3.7%.

That means it cost it 63% to buy products(on credit like terms), and 33.3% to fullfil orders.

So from a risk terms we can say their margin is equivalent to 3.7/33.3 = about 10%. Which isn't bad.

Furthermore, a lot of those costs are variable costs, like shipping, which go down if demand drops.

Amazon has massive control over their margins. Therefore in a downturn, they just turn a dial and are profitable again.

They're just choosing to reinvest profits into growth, and can stop doing that whenever they want.

I think we mostly agree, but not about how much risk they're exposing themselves to in a downturn.

ANd what better what to obfuscate actual company footing - and real power than to prevent people from looking at, analyzing or even knowing what level of error margins you're capable of floating in a downturn.

What better way to additionally hedge against that than to have a many-billion-dollar contract for hosting with the government.

Exactly. Your comment is the most astute so far in this thread.

Amazon is a cash cow that simply feeds a relatively decentralized innovation engine. Moreover, the strategic moves they make (intentionally keeping GAAP margins thin, not separately disclosing AWS in the financial statements for years, etc.) should always be assumed to come from multiple, well-considered motivations.

Someone says, "It's just customer-obsession." Yes, that's the simple myth/story they want you to believe.

That is of course looking only at Net margins. Assuming they are actually on thin margins from there Gross Margins which is still a respectable 30+%. Amazon does go into very deep to try and hide their net margins or competitive advantage. They continue to invest and optimise their operation.

Sure but who will have lower margins than Amazon in a downturn; also higher margins also means more competition from people with lower margins so I can’t really agree with your analysis.

> Jeff Bezos is very focused on this “absolute dollar free cash flow metric.

I wish Apple operated that way, as lowering Apple device price will very likely increase the absolute amount of dollars coming into their back acct.

I mean I get Amazon and Apple ear their income via different methods but still.

But Apple is a luxury and lifestyle brand. And as good as their hardware is, it's not particularly hard to replicate, especially if they were operating on the same profit margins as everybody else. Apple has to zealously maintain their brand, which means limiting both supply and product lineup. Otherwise they may as well change their name to Sony.

Amazon has an enormous lead in scale. More importantly, the only way to defend that lead is to consume as much market demand as they can, which puts downward pressure on their prices.

Too much market share would be the death of Apple. Too little market share would be the death of Amazon.

There processor design technology is ahead of the rest of the industry. The iPhone 6s from 2015 was faster than all 2018 Android phones in single core performance.

When you go down to processors like the ones in the Apple Watch or the AirPods, no one comes close to the size/power. There is a reason all Android watches are big ugly monstrosities.

You realize as far as marketshare, iOS has close to 50% in the US and well over 30% in most other industrialized nations. There is no exclusivity to owning an iPhone in developed markets.

I had thought Apple's smartphone market share peaked in the 20% range. I stand corrected.

I’m only referring to “industrialized”/“developed” countries like the US, Japan, Western Europe, etc.


Benchmarks don't matter. They're both snappy and useable. Android has far more features and variety, though.

Watches are dying. The reason Android isn't trying is because it's a passing fad.

The mid range Moto G that my son had was far from snappy and that was usually hailed as a good midrange Android phone.

But now you sound like Mac users back in the Mac 68K and PPC days.

Apple Watch sales have increased not decreased.

You're comparing a Moto g to an iPhone? That's double the price.

I’m comparing a a 2018 Moto G to a 2015 iPhone.

The iPhone from 2015 will probably be getting updates longer than the 2018 Moto G.

Again, you are comparing a phone that was half the price at launch. Operating systems evolve. Just because a phone is newer, doesn't make it snappier. There are far more things running on Android and iOS by default than there was 8 years ago. Try comparing the 2015 iphone to a galaxy s6.

Here is a benchmark comparison.


Both phones were around the same price at launch.

The iPhone 6s is still supported and still getting os upgrades and security patches - the Galaxy S6 isn’t.

The iPhone 6s should get at least one more year of upgrades but considering the iPhone 5s is two years older and still getting updates I would bet on at least two more years.

Long term, which is the better but?


Here is a comparison between the flagship Galaxy S9 to a two year iPhone 7.


Again, those benchmarks mean nothing. User perception does. And "updates" are a farce. What kind of updates are you waiting for from Apple? Is there a huge feature that you just can't live without?

Btw, your link shows how much better the Galaxy phone is. But if you choose your phone based on which has higher benchmarks rather than things like a better camera, to each their own.

Android is not exactly known for its buttery smooth interface...



Not to mention all of the security vulnerabilities that don’t get patched.

The advantage of updates is that it encourages app developers to take advantage of the latest features. For instance, the author of Overcast, my favorite podcast app could release a new version that required iOS 12 a month after it came out without having to lose any customers - anyone who could run iOS 11 could run iOS 12.

As far as updates, just the integration of shortcuts and Workflow has done wonders in iOS 12.

Link to your list of unpatched security vulnerabilities please.

And apple lags just as much.


Shortcuts is a weaker version of Tasker. You could have switched to Android and had that 5 years ago.

It’s not that Google doesn’t patch vulnerabilities, is that between the phone manufacturers and the OEMs that the patches sometimes never reach the older phones or sometimes no update reaches the phone.

As far as “batterygate”. Batteries degrade over time. Either Apple slowed the phone down or the phone shuts off.

You could’ve just said non-Intel.

> Benchmarks don't matter.

They Do, up to a certain level, and we haven't quite reach there yet across all spectrum of devices.

> They're both snappy and useable.

They are not.

>Watches are dying. The reason Android isn't trying is because it's a passing fad.

The data tells a completely different story.

A shame Apple is acting in a way that makes them earn less. If only they would listen to you instead of that Tim Cook guy! /s

>I wish Apple operated that way, as lowering Apple device price will very likely increase the absolute amount of dollars coming into their back acct.

I've always thought they should make a $10,000+ model. iPhones are essentially a status symbol unless you buy it for privacy purposes. There is a decent market who would pay a lot of money to have a iPhone that the plebes can't afford. Hell, even a lot of broke people would save up to buy one

It's like cars, if you just want a tool to get from point A to B you can buy one cheap. On the other hand you can also spend hundreds of thousands on luxury cars.

Make some limited edition iPhones every year and I'm sure there would be demand and an eventual collector market

How is anything a “status symbol” with over 30%-50% market share in industrialized nations?

There are a lot more reasons to buy iPhones than status symbols and privacy. Chief of which is better performance and longevity.

The iPhone 5s from 2013 is still getting OS updates. The iPhone 6s from 2015 will still outperform most new Android phones and should be getting updates for at least another two years.

>How is anything a “status symbol” with over 30%-50% market share in industrialized nations?

That's my point, when it came out the marketing around the iPhone was how using apple made you a trendsetter and you were ahead of the curve, now everybody has one. So raise the price and bring back the sense of exclusivity

as for performance, it really doesn't matter for luxury items. People pay hundreds of thousands of dollars for old cars and it's not because they perform better than a modern economy class vehicle

It actually didn’t. This was the first iPhone ad.


Or you can watch every single iPhone ad. None of them scream exclusivity.


There are other iPhones for sale besides the $1500 maxed out iPhone XS Max.

They tried this with the watch. There was a $17k version at launch, which later was downsized multiple times until it was eventually killed entirely in 2018. https://www.theverge.com/circuitbreaker/2018/9/12/17851918/a...

Company executives are often smart, and often know their business better than anyone else, but they're also biased. What comes out of a CEO's mouth is very likely to have more to do with what he or she wants _you_ to think than with what _they_ actually think. So taking CEO's grand theories with a grain of salt is in general a good idea for investors.

As to why margins matter... Bezos' quote talks about how valuations are multiples of cash flows. This is true, but multiples aren't constants. Basic investment theory suggests that the multiple should be an increasing function of the company's growth rate, its ratio of sales to assets, and its margins. So if you want to maximize your company's value you ought to be concerned with all of those things. Which isn't to say that trading lower margins for higher growth is a bad decision in Amazon's case, but it's not like it's an irrelevant number.

When you don’t have a standard means of describing a company’s numbers like GAAP you end up with something like what WeWork does.


However, in its disclosures to bond investors, the company added back items like stock-based compensation, sales and marketing expenses, even general and administrative expenses to arrive at what it called a “community-adjusted EBITDA” of positive $233 million.

Great read. Points #5 (on being able to use a longer time horizon when investing in ideas to let them grow) and #6 (on the growing importance of consumers) stood out to me as well.

> People really need to get off of the "income analysis" trope with Amazon.

What does this have to do with the article? Percentage margins aren't mentioned anywhere. It mostly talks about absolute dollar operating income which, unlike GAAP income, isn't too different from free cash flow. For example: "Operating income for AWS was $7.29 billion in 2018, up from $4.31 billion in 2017. [...] And AWS is the lion's share of Amazon's operating income: Nearly 58% in the fourth quarter of 2018, and nearly 59% for the full year." So AWS is certainly a major driver for Amazon's free cash flow.

Because GAAP is a useful tool for analyzing a business, independent of what the CEO might say. The CEO is not exactly an objective observer to his own company.

Businesses are judged by the income they produce because the reason businesses exist is to generate income. If people want to run a charity they would presumably register as a charity. There are interesting edge cases, but the No 1 company on the NASDAQ by market cap is not one of them.

"It's not how they run their business" is an interesting observation, but businesses ultimately don't get to choose the yardstick they are measured by. In the long term, that is income.

I do wonder what is going to happen when/if there is more margin compression in the cloud space. Specifically when it comes to Amazon, because Amazon really has razor thin margins in its retail sector. Other cloud providers have high profit margins outside of the cloud (Google, Microsoft, Apple) so they could still be ok. They all have so much pricing power too, that I wonder if there is some mutually assured destruction of margin that prevents them all from competing heavily on price. Or if the space is growing so fast that they simply don't need to compete on price as much.

Amazon has people locked into Lambda, Google and Apple have people locked into device integration, and Microsoft has people locked into Office and in a short while Github CI integration.

All three are trying to differentiate more and more (with varying degrees of success) while they increase the price of the commodity services so it doesn't compete with the differentiated ones. They will not compete on price with each other.

I just don't know what will happen if somebody plays an Amazon at the commodity segment, and go selling services with thin margins.

I've always wondered if FB could get into the cloud game, it seems like they might have some data center expertise.

However, your comment has me thinking that FB doesn't have a way to lock people to their cloud - they'd be forced to compete on price more than any other major cloud provider. Probably not a great business.

But is Apple a cloud provider? I recall they have some data centers but I thought that it was rumored they used Google Cloud or AWS (or both) too.

Internally, Facebook has always struggled with infrastructure capacity throughout their history. They still have growing pains as far as physical space for equipment goes.

When Amazon got into the cloud business, they actually had most all of the non-software challenges solved already. It turns out that datacenters and distribution centers go in the same kinds of places and have very similar kinds of challenges. They benefited significantly from knock-on effects from running their existing business.

If Facebook got into the cloud game, they would have to learn two new businesses at once: cloud hosting and industrial real estate.

I don't think that they would be able to obtain good pricing on either real estate or server hardware. Their current hardware vendor can't absorb that large of a scale out.

> When Amazon got into the cloud business, they actually had most all of the non-software challenges solved already. It turns out that datacenters and distribution centers go in the same kinds of places and have very similar kinds of challenges. They benefited significantly from knock-on effects from running their existing business.

That sounds good in a blurb in CIO Magazine but is likely completely disconnected from reality.

The complexity of building a cloud provider is not primarily in the physical management of physical assets in a big box building.

Having worked in several industries before landing in software development, I can tell you that the most common form of hubris we have as an industry is in thinking that our domain is where the hardest problems live. Or in thinking that the level of difficulty of our challenges has a 1:1 correlation with those of the business.

Building data centers is hard because there are not a lot of places that you can build them effectively. Large scale, specialty real estate deals are the kind of thing that effect a company's financials in a big way and for a long time -- much longer than the market cycles where you determine whether or not to continue or abandon your hypothetical, nascent cloud offering service -- and are exactly the type of thing that market investors will pillory your company for if you fuck up.

If you're Amazon and you're starting off in the datacenter game and it doesn't go well, you can always transition the property into a distribution center. Those are useful to you anyway.

If you're Facebook, what are you going to do? Who are you going to sell this $800-1200/sq ft (that is the cost of building a turnkey DC and puts AMZN's largest datacenter somewhere between 172-258mil and that's just one of them) "big box" to? They should make a multi-billion dollar investment for that and hire a global enterprise sales organization on top of that to sell it?

Maybe it's not so much that what CIOs say is divorced from reality but that they see reality at a scale that you don't. Their abstractions may not make sense to you.

microsoft did it pretty easily, albeit 4 years later than amazon, which is partially why they're playing catch up. microsoft didn't have any "industrial warehouse buying" expertise and was a lot more familiar with server tech than amazon was at the time they started aws in 2006, and probably had a good number of data centers already, they just weren't built for rental services.

although it's outside facebooks current core competency i think they wouldn't have too many issues getting into the cloud game if they wanted to. they could hire all the expertise they need and get 100 data centers up and running in 5 years.

but why would they? it's like google getting into the ISP market to compete with verizon et al. google runs its very profitable business on top of the commodity rails and keeps limited deployments of google fiber as a stick to threaten slow or uncooperative ISPs. facebook can run its profitable software based businesses on top of the infrastructure instead of joining the cutthroat race to the bottom of cloud computing that may or may not be semi nationalized/regulated as ISPs and utilities are today

Microsoft did it pretty easily because they have the financial flexibility to make a multi-billion dollar investment and already have a global enterprise sales and support organization with a 20 year head start on the competition.

Facebook can't say the same and getting 100 datacenters up and running would cost them 25 billion before hiring anyone -- that's more than half of their cash.

I agree with you though that it makes zero sense for Facebook to do this.

Another consideration is that FB would have to reorient their culture to sell to and service the enterprise customer. You could argue that Google is having that very problem with their cloud offering. Despite having superior engineering and product, it’s just not in their DNA to cater to the enterprise in the same way it is for Amazon and Microsoft. FB might have considered cloud a few years ago, but I think they wisely stayed out to focus on their core offerings.

Is Google cloud a superior product with superior engineering?

As an enterprise customer of all three, I will tell you that both Google and Microsoft have terrible UI, terrible permissions sets and terrible documentation. Microsoft has terrible policies and support but their interface still edged out Google. Google constantly shutting off APIs and forcing people to refactor isn't doing them any favors either.

IAM and the documentation are AWS' secret sauce. It's the thing that wins over those internal champions/coaches that an enterprise sales organization needs at their prospect companies to close large scale deals. In an environment where I have to manage security, compliance and administration on top of just using the thing, it makes my job 1000x easier.

They pursued that for a short while when they had self doubts in their ad business. This resulted in them buying Parse. Only later to find out their ad business is solid and shutting down parse.

> FB could get into the cloud game

They tried that with Parse acquisition but shut it down

I don't think Lambda-lock-in is nearly as powerful or widespread as OS/device. Most frameworks these days are container independent.

Microsoft also has Active Directory and Visual Studio.

> Specifically when it comes to Amazon, because Amazon really has razor thin margins in its retail sector

What do you consider thin margins? Their overall margins are better than Walmart's.

Amazon: 65_932_000 / 177_866_000 = 37% [1]

Walmart: 126_947_000 / 500_343_000 = 25% [2]

[1] https://finance.yahoo.com/quote/AMZN/financials?p=AMZN

[2] https://finance.yahoo.com/quote/WMT/financials?p=WMT

Just looking at Q4 results [0], looks like operating margins for the businesses are:

US Retail: 5% [$2,251 / $44,124] Intl Retail: -3% [-$642 / $20,849] AWS: 29% [$2,177 / $7,430]

Really striking diff between Retail and AWS biz.

Walmart, for what its worth, had operating income of $20B on $500B in revenue, so about 5% there too.

[0] https://ir.aboutamazon.com/static-files/a5035fcd-5646-45df-b...

Operating margin is not Gross Profit/Total Revenue, its Operating Profit/Total Revenue. You were not surprised by the number 37% you got?

That's the gross margins for ALL of Amazon's business, so the fat profits of cloud are included. Their margin on physical goods must be lower.

Apple? Did I miss something?

They double and triple dip customers pretty hard today, I'd imagine margin lost on basic services like EC2 would shift to AWS specific services they can layer more costs on top of (i.e. RDS)

> Amazon really has razor thin margins in its retail sector

Not any more, if their advertising business keeps skyrocketing, Amazon will be US's Alibaba.

When you’ve got Kubernetes on top, it literally makes zero difference who the underlying cloud provider is.

And thank god Kubernetes exists. Not about to waste my time relearning the API of every new cloud offering a company wants me to try.

Sort of. It depends if you want to use Kubernetes to operate your own S3, your own RDS, your own SQS, your own Kinesis, your own Cloudfront, your own DynamoDB, etc.

If you have the dev resources to do that, great. But lots of other players will prefer to use off-the-shelf stuff from cloud providers.

I think that’s a different question you’re answering though. For sure, continued automation is 100% the modus operandi. I wouldn’t be surprised if in the future IT departments don’t really exist anymore.

But even with “off-the-shelf stuff”, your site reliability engineers would still be using Kubernetes as the primary language to communicate. Or I guess you could go all cutting edge and go one layer up with Istio instead. In that case then, Kubernetes is still there hiding underneath.

Until Amazon, Microsoft, or Google move up the stack.

(Or maybe all three. They really are eating the industry, but they add value so, yeah, I guess they deserve their rent. Just sucks for anyone else in that space.)

I think it’s inevitable. Cloud providers will continue to move up the stack so other firms can focus more resources on their unique and differentiated offerings. It just seems like in the future most companies shouldn’t have any notion of servers/containers etc, they will only focus on their core business logic. We’ll just keep going upupupupup.

Ever upwards abstraction my friend. Seriously, I thought Kubernetes would be it, but now all my coworkers are trying to get me to use Istio, which at first glance seems to be ... a managed containerization for different Kubernetes managed containers?

Turtles all the way down I guess, which now I’ll have to spend all weekend on researching unfortunately.

If I’m using a cloud provider product like text-to-speech then which part of Kubernetes abstracts those APIs so I can move from Polly to Bing with no work? I didn’t think Kubernetes covered things like that?

In short, Kubernetes does not abstract any APIs. There are some ways to use things like Terraform to deploy and run in different cloud environments, but those API have different features that are inherrent and not transferrable.

Terraform doesn’t abstract creating resources on different cloud providers. Each provisioner is tied to specific resources for the specific cloud provider.

Kubernetes doesn’t help with storage integration with S3, ETL processes involving Redshift, using queues and notifications setting up security, etc.

Kubernetes is no more the be all end all of maintaining cloud agnosticism than using the repository pattern means that you can tell your CTO to get rid of your company’s multi million dollar Oracle installation.

so, with operational margin 28.42%, aws outruns apple with operational margin 22.72%.

if you've told me aws takes more premium on their products than apple per $, i wouldn't believed.

You should compare AWS's pricing vs dedicated providers, it's near a 3x difference compared to OVH last time I looked.

The reason they can get away with this is because they make it so difficult to figure out the total price.

Yep. A lot of the cost is hidden in their bandwidth costs, it's extremely overpriced.

US retail also drives more than half of Amazon's operating income.

For Amazon, the cloud is the little engine that could. Amazon Web Services comprised just 11% of the company's overall sales in 2018, but delivered more operating income than all other business units combined.

I think that's the point the article is trying to make (first sentence of the article)

Wait a second....

It's surprising, but correct.

In 2018-Q4, North American retail drove $2.25B (59%) of OI; AWS drove $2.18B (58%) of OI. International retail lost ($642M) (-17%) of OI.


Don't see much discussion around AWS support. Man, In my opinion its not AWS is great, its more about GCP & Azure are not even good.

Do you work for a company that has a business support plan?

AWS is a byproduct of Amazon's e-commerce business, that became a viable service of it's own. It was a brilliant move to monetize the excess compute power Amazon had to have on hand for it's e-commerce business. Without the e-commerce business, AWS wouldn't exist.

This is a myth, and it needs to die. AWS was not designed for amazon's infrastructure, nor was it spare capacity of said infrastructure at any point. It was built from the ground up to be an independent service selling to third parties, on independent servers in independent data centers.

It took many years for amazon.com retail stuff to migrate to it, and supposedly some parts still aren't migrated.

I mean, I'm gullible so I believe you, but I also believe Steve Yegge who wrote about this:

> Well, the first big thing Bezos realized is that the infrastructure they'd built for selling and shipping books and sundry could be transformed an excellent repurposable computing platform. So now they have the Amazon Elastic Compute Cloud, and the Amazon Elastic MapReduce, and the Amazon Relational Database Service, and a whole passel' o' other services browsable at aws.amazon.com. These services host the backends for some pretty successful companies, reddit being my personal favorite of the bunch.


(posted in 2011, and he claims he left amazon 6½y before that)

You sound confident so I'm sure there's something to it. But on the other hand, there's loads more, similar comments in that memo.

And Steve, well, Steve is known to be quite the iconoclast. I never took him to suffer fools gladly.

Both of you can be right. It's possible that Bezos saw the opportunity in taking advantage of their tech infra expertise and building that into a public product, but maybe it wasn't just taking what they already built and making it public.

I don't personally know but https://www.networkworld.com/article/2891297/cloud-computing... makes some strong claims otherwise.

And if AWS were just Amazon's spare capacity, who would bet on their stuff staying up through cyber monday or prime day?

Nobody is claiming that it's still just spare capacity, just that it started as that.

I love this:

    "Bezos is super smart; don't get me wrong.

    He just makes ordinary control freaks look like stoned hippies."

Andy Jesse wrote the original business plan for AWS. He says something different.


Is there a source for this? I'm genuinely curious. The "AWS started as Amazon's extra capacity" is such a compelling story.

It wasn't extra capacity so much as making some of the internal tools publicly available, and it took off.

I believe SQS was the first publicly available service. EC2 was obviously the tipping point and that was a couple of years later.

The story I always heard was they had a ton of extra capacity that was very seasonal and then someone came up with the great idea to sell that extra capacity in the off season.

SQS as the first publicly available service sounds weird to me but I don't have any sources to cite otherwise.

Edit: Wikipedia says SQS was the first service https://en.wikipedia.org/wiki/Amazon_Web_Services#History

I recall S3 being the first product announcement, with no real mention of what was yet to come.

SQS came before S3.

Part of that success was a mandate back in the earlier days of the company that all development and systems administration assets be designed with the presumption that they may be converted into a customer accessible frameworks one day.

I've never heard this before, it's an interesting anecdote, do you have a source?

Mackenzie Bezos now drives half of Jeff's income

I'm a recent convert to 'Boglehead' investing. Among the principles Bogleheads follow, we don't like to pick individual stocks-- index funds are far safer.

But I cheat a little. One of my small holdings is Amazon. They look set to dominate in multiple arenas. Fingers crossed at least one pans out!

AWS, Facebook and amazon are going to get absolutely crushed if there’s a recession that triggers a tech industry pullback. All the VC funding that goes to these hot tech companies goes to Facebook and google for ads and AWS for server space.

In addition, kubernetes is going to start squeezing aws’s profit margins, particularly ec2 as people move to commodity k8s hosting.

Is the implication that people/businesses will stop developing and using web services? I just don't see how we can go back now. Further, web tech is cheaper to scale than anything else, so when the wallet tightens it seems that's where budgets would focus.

Also, AWS moves faster than web tech on the whole, so even if the general shift to k8s continues, AWS will be there to make it easier than rolling your own, and at a price point that's "worth it."

Historically, teams would just make their existing servers last longer. Considering most of AWS premium services have an open source alternative, guessing most people will port to containers. Built on open source has been their source of strength but is also their achilles heel. They shortened the learning curve on to their platform and off of it.

That's a complete no-sense. When you buy from a public cloud you buy reliability, availability, security and support.

The service layer is just an add-on. Part of their premium. Even if you go full open source you still need infrastructure to run it and that's exactly what companies don't want to do and won't do anymore, and the reason why AWS and Azure are massively successful.

I think I must have miscommunicated my position. I'm not reflecting on cloud as whole, but how AWS is particularly vulnerable to a tech sector downturn. In the event of economic downturn, existing cloud customers will optimized costs (pre-cloud they would have avoided infrastructure upgrades, now they will focus on different cost structures). My belief is that generic services, such as storage, compute, ect. will priced like commodities because the market is more efficient since multiple cloud provider have the same service. Tools like terraform allow easy enough cloud configuration porting that cloud providers have little pricing power for these products. One area of cost optimization is to utilize generic cloud resources instead of AWS specific services (where I see many of my costs). Many of the services that generate positive cashflow for AWS seem to be built on top of open source which means that converting existing costly services can be converted to use generic resources with out too much implementation difficulty.

I'm only a developer so I'm sure I'm missing Operations and Business perspectives. Would love to learn more.

Those positive cash flow services are managed. It’s cheaper to run your own Postgres instance in EC2 than it is to run RDS, until you have to pay someone to operationalize it.

You pay $.09 an hour more for an RDS m5.large than you do for an EC2 m5.large. What kind of DBA can you get for $67 a month? That’s maybe one hour of a good DBA’s time per month, tops.

Now, someone will correctly point out that you can run way more compute on a colo’d server than an EC2 instance and they’d be right. But when you realize you’re paying AWS for hypervisor patching, network automation - I’ve worked in places where a security group change in AWS was equivalent to a two week servicenow ticket with the network team - ...

The business perspective to AWS is that you bake operational costs into the product and stop paying people that generate those costs. Not good for sysadmins who aren’t willing to change, but it’s where we are. And it’s why you can’t just say “open source makes this convertible without difficulty” - somebody’s still gotta manage those services ;)

The myth that Teraform makes cloud migration easy needs to die.

Have you looked at the different TF provisioners? They are all provider specific.

There are a lot of services on AWS that are specific to AWS that you would need to rearchitect your system to use. Even something as simple as an ETL process that lets you load directly from S3 into Redshift and Aurora.

Sorry, how is k8s getting more widespread cut into ec2 nodes? (if anything k8s needs a bit more nodes)

k8s allows for more complete utilization of nodes, and scaling out smaller cheaper nodes instead of scaling up bigger more expensive ones

Do you have any link handy to elaborate on this? Because when I create a K8s cluster I define their size and number (and autoscale, same as with AWS autoscale), I don’t know how k8s can spin up smaller nodes in a different way than say, AWS

I'm not the original poster, but I think they were suggesting that K8s supports better bin packing than you get with virtual machines, so you can run the same applications on fewer (or smaller) servers.

In terms of the cutting edge of bringing "declarative, Kubernetes-style APIs to cluster creation, configuration, and management" see https://github.com/kubernetes-sigs/cluster-api

There is a lot more to AWS than just EC2.

Large companies want to move their databases storage and compute off prem ASAP. Any tech/economy pullback should only accelerate that.

It's too easy to steal market share with software. There are no moats. Kubernetes is the first of many equalisers, crossplane.io is next. There's no monopoly effect.

That's an absolute fantasy. I've worked at companies who tried to move between cloud providers.

Unless you build your service like that from the start (e.g.: prematurely optimize, which practically no successful company has the chance to do), you will experience some amount of AWS lock-in.

Applications are open for YC Summer 2019

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact