Imagine how little fun online retailer Amazon would be having on its quarterly calls if it had not launched its Amazon Web Services cloud almost eleven years ago. The very premise of Amazon was to eliminate brick and mortar retailing, cutting out capital expenses as much as possible, to deliver books and then myriad other things to our doorsteps.
How ironic is it that Amazon pivoted to one of the most capital intensive businesses on earth – running datacenters – and has been able to extract predictable and sizable profits from it to prop up its other businesses and strengthen them.
It is a remarkable thing, really. That said, transforming the IT sector, which Amazon Web Services, its public cloud computing arm, has truly been doing with a joyful ferocity, is not an easy or predictable thing, and with Google, Microsoft, IBM, and myriad upstarts as its competition, Amazon is the juicy target everyone is now gunning for. And despite the growing demands of now millions of customers – Darrin Manney, head of investor relations at Amazon said “millions of active customers,” that’s plural, on a conference call with Wall Street analysts going over the company’s financials for the fourth quarter of 2016 – and the diversity of services it is offering, Amazon is still growing at a healthy clip for an IT supplier in a relatively flat market.
But, that growth is slowing, and the Amazon business as it is currently configured – sells a mix of raw infrastructure and cooked platform services – is finding its own level in the IT sector. This is natural enough for all IT suppliers, which triple their revenues when they start going up the hockey stick curve, than then they double revenues for a while, then it cools down to maybe 80 percent growth and then 50 percent growth and then 25 percent growth and then the rate of change gradually slows until it matches the overall rate of spending in the IT sector overall. Unless and until the company launches a new and different product that has explosive growth, and if a vendor does that, it is often a product that cannibalizes their own products (if they are lucky) and accelerates the slowing of the growth rate, and if they are not lucky, someone else does it and eats their lunch.
We are a long way from that point with Amazon Web Services, and the $14 billion annual run rate that the Amazon public cloud is now humming along at is a mere fraction of the $3 trillion IT and communications budget for all of the corporations, governments, and academic institutions of the world added up. We have a hard time imagining AWS growing to be as large as the Amazon retail business, but that is indeed the company’s goal, as it has said many times, and it will be interesting to see how the top brass at AWS pull that off. A profound and deep recession might help. We tend to have them every decade or so. It has been almost a decade, and there is no shortage of chaos in the world as best as we can see.
In the quarter ended in December, revenues for AWS were up 45.8 percent to $3.51 billion, and operating profits rose by a much larger 59.7 percent to $926 million. For the full year, AWS had $12.19 billion in revenues, up 54.7 percent, and operating income of $3.11 billion, more than double and comprising 25.5 percent of revenue. That is the kind of margin that enterprise software companies are able to take down if they are healthy and strong with a large, committed customer base, and this is really quite an achievement. It is one thing to take glass houses wrapped around servers and storage and switching and make it consumable like software. It is quite another thing indeed to make it perform financially like software.
The uptick in operating profits from AWS was done despite price cuts on numerous products that kicked in on December 1 last year and despite a large spike in investments – both direct acquisition and leased properties and IT capacity – for the quarter. Take a look:
The chart above shows all capital investments that Amazon makes, including for its massive warehouses for its retail operations. But the general consensus is that the largest portion of Amazon’s capital expenditures is for datacenters and software for AWS datacenters, not warehouses and not software for its own use. As you can see, someone who is selling leases is making some coin, because Amazon had a big spike in capital leases in the quarter, driving its overall capital expenses up 59 percent to $4.46 billion and representing the biggest jump we have seen since 2014. We don’t think that Amazon has its hands on production quantities of Intel “Skylake” Xeon processors as yet, which are due to be formally announced sometime around June or July this year if the rumors are right. But for all we know Intel is quietly shipping them in volume to Amazon, Google, and Microsoft as well as a handful of marquee HPC centers. While this spending rate is certainly high, it is nowhere as high, in terms of percentage of growth, as Amazon did during 2013, where it basically doubled capital spending as AWS really took off in the enterprise. Spending took a bit of a pause in 2015, as AWS techies learned some tricks about squeezing more time out of their gear, and that helped the overall profitability of AWS and the company as a whole.
We are certain that AWS has other tricks up its sleeve to make the most of its investments. Since 2007, when AWS started to become material, Amazon has ponied up $48.4 billion for capital investments, and it has turned that into over $32.1 billion in revenues and $6.1 billion in operating profits. A third of those revenues and half of those profits came from the past 12 months. Given all of the flitting on and off of capacity across millions of customers, creating a machine that is even this predictable and profitable is akin to what IBM was able to do to transform itself and create the IT sector in the first place with the System/360 mainframe.
What will Amazon Web Services do for a second act? While serverless architectures based on services like its own Lambda product may be the wave of the future, that is still, at its core, an IT services model. It would be interesting to see Amazon actually start producing algorithms and application software on behalf of customers, moving itself even further up the stack and deeper into IT departments. IBM used to write software for companies that bought mainframes, way back in the day, and it created industry-specific lines of software that it sold on its mainframe and proprietary midrange iron when computing matured a bit. One could argue the dumbest thing IBM ever did was sell off those software businesses in the 1990s, and one could also argue that the next logical step for AWS is to start creating application software directly for customers.