|AMZN operating margins|
Amazon’s $52.9 billion of revenue in the second quarter of the year came in a tad below what Wall Street analysts expected — and that doesn’t matter whatsoever.Below the fold, I discuss one of the implications of these amazing margins.
That’s because the massive online retailer once again posted its largest quarterly profit in history — $2.5 billion for the quarter — on the back of two businesses that were afterthoughts just a few years ago: Amazon Web Services, its cloud computing unit, as well as its fast-growing advertising business.
AWS revenue growth accelerated in the second quarter, rising 49 percent year over year.A business growing nearly 50%/yr with 25%+ margins is in an amazingly strong position.
The advocates of decentralized storage naively imagine that they can compete with the S3 part of the AWS juggernaut. Their product doesn't perform as well, isn't integrated with a cloud computing environment, is more expensive to implement and operate, and lacks economies of scale.
What if we imagine all these disadvantages are magically eliminated? Del Rey and Molla point out that:
An Amazon that is posting growing profits from its non-core business means an Amazon that can continue to keep prices low and invest in ever-speedier delivery times to widen its defensive moat in its main retail business.The margins on AWS, averaging 24.75% over the last twelve quarters, are what enables Amazon to run the US retail business averaging under 3% margin and the international business averaging -3.7% margin over the same period.
AWS is the source of the investments that drive Amazon's takeover of retailing. So Amazon's "slow AI" would view a threat to disrupt AWS, or even just the S3 part, as an existential threat, and respond accordingly. It has over $30B in cash to fuel the response, and the infrastructure to out-compete a storage network's nodes, driving the price down to make it uneconomic for anyone else to run a node.
This is why I say that It Isn't About The Technology. Competing with the FAANG is primarily a business problem, not a technology problem. Discussing the technology without knowing the business strategy it is intended to support is a chimera bombinating in a vacuum (Hat tip to Rudyard Kipling).
I know that this sounds countrarian, but there are little to no "economies of scale" in general, and least of all in computing, especially "cloud computing", beyond a pretty low threshold.
The reason is that there are instead economies of specialization, and scale often enables specialization, but by itself does not offer much. Sure, a huge metal press costing $1m does stamp parts a lot faster and cheaper than a hammer, but that's because it does only a much more limited range of things than a hammer, and adding a second one does not give any further economies.
For computing "economies of specialization" are already achieved by chipmakers and disk makers and RAM makers, and software makers, e.g. the intel monoculture in desktop CPUs, a very narrow specialization.
The extreme degree of specialization these makers enjoy means low cost of their products, but also that their products must be extremely generic in use. An intel CPU is averagely good for everything, so is a RAM stick, or MS-Windows.
Therefore the users of "industry standard" computing parts cannot themselves economies of specialization; these have been fully used by the parts makers.
For "cloud computing" this is even more true, because "cloud services" must be enormously generic, and that prevents economies of specialization.
The key to the profitability of AWS is that it is very expensive, as in this example:
«Proofpoint rents about 2,000 servers from Amazon Web Services (AWS), Amazon.com’s cloud arm, and paid more than $10 million in 2016, double its 2015 outlay. “Amazon Web Services was the largest ungoverned item on the company’s budget,” Sutton says, meaning no one had to approve the cloud expenses.»
That's $5,000 a year per VM, a crazily expensive amount. For $10m a year you can have a dedicated data centre or two with way more than 2,000 servers. A pretty decent dedicated hardware server rents for less than $1,000 per year.
People pay very high prices for AWS because "nobody ever got fired for virtualizing on AWS" and because it is often an "ungoverned item", and also because in some companies it is advantageous to turn capital spending into recurrent spending.
There are very significant economies of scale:
- Amazon buys hardware much cheaper than smaller operations, even one as big as Backblaze.
- Amazon writes and maintains its own software and amortizes it across its huge customer base. You can't do AWS as efficiently or as well as Amazon on FOSS, or even on COTS.
- Amazon amortizes the operations staff across a huge customer base.
- Amazon does Intel fab style data center replication, to support (for example) the triple replication of S3. You can't do this unless you're buying data centers in bulk.
And your estimates of server cost don't account for overheads such as staffing, power, A/C, network bandwidth (Amazon runs its own network and peers with major backbone providers), ...
But you are right that transforming capex into opex is important for AWS customers, whose cost of capital is much higher than Amazon's. Another big value customers get from AWS is incremental scaling to meet demand, eliminating over-provisioning. Doing it yourself means predicting and provisioning for the peak, which is bad if you under-estimate the peak and expensive at non-peak times when capacity is idle (i.e. almost all the time)
Two illustrations from of the advantages AWS reaps from economies of scale.
1) Chris Williams reports that:
"Amazon has designed its own 64-bit Arm server processors, dubbed Graviton, and is right now renting them out on AWS. ... Amazon claims A1 instances are up to 45 per cent cheaper than their x86 virtual machines, depending on the configuration."
2) Richard Chirgwin reports that:
"Earlier this month we covered ThousandEyes' report into cloud networking performance, in which the metrics outfit observed that Amazon offloads cloud traffic from its network at the first opportunity, compared to competitors like Google.
As Bezos' crew announced just a few hours ago, there's now a way to avoid that – just pay Amazon to keep traffic on-cloud.
That's the premise of the AWS Global Accelerator: rather than handing traffic off to a generic internet connection, customers can get AWS to carry the traffic for them.
"Traffic routed through Accelerator traverses the well monitored, congestion free, redundant AWS global network (instead of the public internet)," Amazon explained."
Post a Comment