Amazon Web Services is a big business that is growing fast. Amazon has typically remained quiet on the subject of AWS financials, but has finally opened up about how the service is doing. CIO.com columnist Bernard Golden looks at what AWS revenues really mean.
The importance of this announcement is that it reveals what has been intuitively obvious to most observers: AWS is a big business and growing fast. For the first time, the notoriously secretive Amazon has opened up about AWS financials and given everyone a look at how the service is doing. Of course there were a great number of articles (some linked to in this piece) discussing the revenue announcement implications. Most of them were in the vein of a horse race — how far in the lead is AWS, how fast are the other horses, will AWS flag in the stretch run, etc.
None of them, however, evaluated what AWS’s numbers mean in terms of the IT industry and overall application of information technology in our society and economy — and I think that’s where the real meaning of AWS’s revenues emerges.
AWS revenues today and in the future
In late 2014 I projected AWS revenues through 2020. In that piece I gathered estimates from a number of sources, including the much-lamented GigaOM and various investment banks; the consensus of those estimates was that 2014 would see around $5 billion and around $7 billion in 2015 (pretty close to AWS’s actual results).
I went on and extrapolated growth projections from Pacific Crest (it said AWS was growing at around 40 percent) and estimated that AWS would see $6 billion in 2014 and around $9 billion in 2015; obviously I was wrong on the high side. In that piece I noted that, by using the Pacific Crest growth numbers, AWS would, by 2020, achieve around double the consensus revenue numbers ($40 billion vs. consensus $20 billion).
However, while I was on the high side, it’s important to recognize that the numbers I estimated aren’t wrong — just a few years early. And, given that AWS is growing approximately 25 percent faster than Pacific Crest estimated, it might be that my 2020 numbers will actually end up too low, not too high.
In the figure below I updated the revenue chart from my previous piece to include a third projection based on Amazon’s announced revenue and numbers and its statement that the service is growing at 49 percent. In this third set I have used 49 percent throughout the next five years even though Amazon said AWS’s growth is actually accelerating.
As one can see, this third line shows AWS revenues at around $46 billion in 2020, 15 percent higher than my previous estimate, and evidence of what the impact of a high growth rate can be. It’s surprising that in all of the commentary on the AWS revenue announcement, no one focused on the likely future numbers for the service, which, if the growth rate continues, become truly staggering pretty quickly.
Of course, this raises the question of the AWS growth rate. Naysayers will no doubt voice a belief that AWS’s growth is bound to level off soon. However, to state the obvious, AWS’s growth is based on two things: supply and demand. The next section of this piece will address the demand side, but one can look to AWS’s margins for a clue to the supply side.
The curious thing about these margins is that they’re so un-Amazon like — they put paid to the oft-voiced incumbent vendor lament that AWS prices can be so low because Amazon is ‘subsidized’ by the capital markets, while they have to show profit.
As this Seattle PI piece describes, Amazon’s core business strategy is low prices, expressed via cost efficiency and low margins. This drives volume, which reduces costs, etc., etc. In other words, low margins are baked into the Amazon strategy (the one time Amazon strayed from this, with the Amazon Fire phone, it was unsuccessful and derided; I maintain that if Amazon had followed its core strategy and priced the phone at $99 initially, it would have achieved very different results).
So why has Amazon pursued a different margin strategy for AWS? I believe it’s because AWS is supply-constrained. Amazon is working flat-out just to keep up with current demand; increasing demand by dropping prices (and reducing margins) would be self-defeating, and pose a real business risk — lack of capacity, which could alienate current or potential users.
The question of capacity is quite important, and rather controversial. One insightful commentator, Charles Fitzgerald, says that Amazon needs to spin AWS off so that the service can be free of the company’s seemingly unending new offerings (e.g, Echo, Amazon Business, Dash, which I wrote about here) that compete with it for capital, and can therefore grow even more rapidly. This is in part based on a topic that briefly became hot during the past quarter, Amazon’s use of capital leases, which appeared to negate the company’s ability to drive free cash flow in the face of negative profitability.
This concern about capital leases is misplaced. These leases reflect Amazon using every dollar it can access to invest in its business, which is what companies that view themselves as participating in a transformational shift offering unprecedented opportunity do. Three decades ago the cable industry was criticized for continually borrowing to fund marketing and mergers, rather than building profit margins and paying dividends — and we all know how that turned out. Cable became one of the great businesses of the last decade able to finance (although not complete) $40 billion mergers.
Amazon believes it is the leader in what will, in retrospect, be seen as a seachange in the way we work and shop. It wants to invest as much as possible in that, so it manages for cash flow and is borrowing for even more firepower in its manic devotion to winning this battle.
Why does Amazon manage AWS differently than other business units?
So the conundrum about AWS margins is why they are high, not low as the Amazon business philosophy would dictate. Why is Amazon managing this business so differently than its others?
I believe that the biggest issue for AWS executives is keeping up with customer demand. Two years ago I estimated that Amazon needed to install $6.5 million in equipment each and every day of the year; I may have estimated too high a figure, but it’s clear that Amazon has to manage a tremendous engineering process to provision enough computing resources to meet demand. To state it bluntly: Growing a $6 billion computing infrastructure 49 percent annually is a herculean task.
Therefore, cutting prices (and margins) isn’t in AWS’s interest, given the scope of this task. When you’re working flat-out to keep up with demand, the last thing you want to do is create even more demand. A 49 percent growth rate shows that there is no shortage of demand — and demand is where the really interesting implication of AWS’s revenues displays itself.
Carr’s perspective gained many adherents, and the outsourcing craze of the past decade reflects it. In effect, Carr believes that everything important about IT has already been created, and the remaining challenge is to squeeze cost out of it.
What’s interesting about AWS’s revenues is how little they support this point of view. Overall, it’s clear that the bulk of AWS use is for new applications rather than as a lower-cost alternative to on-premise hosting of legacy applications.
And these applications are far from commodity, me-too offerings. AWS has enabled a plethora of innovative startups, allowing them to create new IT-based businesses that have, in many cases, rapidly grown to be significant in their own right. To name just three:
Netflix makes video consumption an always-on, available-everywhere user experience and is shaking up the established ecosystem of media creators and distributors
Airbnb provides a platform for lodging, disrupting the existing hotel industry by making a far wider range of traveler options available throughout the world
Pinterest lets people create virtual scrapbooks, allowing like-minded enthusiasts to share valued content
It’s no exaggeration to say that AWS has dramatically changed the economics and timescales of starting new companies, and the fertile soil of the service has allowed thousands of new companies to germinate, push up new shoots, and in many cases, flourish.
However, a more recent development is the dramatic growth in AWS enterprise adoption. Each week I encounter enterprise IT groups — many of them in industries one would surmise would be reluctant to use AWS — developing plans to use the service in a major way. Many of these companies are ones that two or three years ago vowed that they would never use AWS.
The motivation for these enterprise initiatives is surprisingly similar to that of the startups — an ability to quickly and cheaply deliver new applications with innovative qualities. In that sense, these initiatives embody Marc Andreessen’s “Software is eating the world” mantra, which I discussed here and here in pieces on the Third Platform. Simply put, far from IT being a commodity, AWS is enabling and benefiting from huge growth in IT innovation.
Therefore, the de facto view on the part of the startups that AWS is their natural computing environment, together with the increasing adoption on the part of enterprise IT, indicates that AWS’s growth rate is likely to continue for the foreseeable future, i.e., the next decade. Indeed, one of the reasons for Fitzgerald’s call for AWS to be spun out of Amazon is his feeling that AWS faces a trillion-dollar opportunity. Baldly stated, demand for AWS is huge, growing, and long-lived.
It might seem that this belief is overblown, but it is exactly in line with other IT platform shifts of the past. Each platform change led to a huge expansion of use and a vastly larger overall spend on IT. It’s likely that AWS will be the beneficiary of a similar phenomenon, as IT expands into more and more segments of our society and economy.
What AWS revenues mean for incumbent vendors
It’s a truism airlines and automobiles supplanted railroads for personal travel because rail operators failed to understand they were in the transportation industry, not the rails and railcars industry. People used railroads as a mechanism to get from one place to another — and once another option was available that was faster and cost-competitive people flocked to it. A mere two decades after the industry’s peak during WWII, it was on its last legs, with many passenger railroads going bankrupt and the remainder abandoning passenger travel nearly everywhere.
Most incumbent vendors view cloud computing as a recasting of infrastructure delivery — an offering that delivers computing and storage with more agility. Therefore, most of them have responded to AWS by creating a cloud offering that incorporates their infrastructure product, claiming that its superiority makes AWS much less compelling; in other words, they are framing the competition on the basis of the quality of the underlying infrastructure. Unfortunately, by doing so they indicate a fundamental misunderstanding of what is driving AWS adoption.
AWS has prospered by making it easier and faster for users to create and deliver applications. It began by offering infrastructure as a service, which, by itself, sparked immense demand. It has now added a panoply of additional services to the core infrastructure offering. For years people have talked about Unix/Linux as a set of Legos — services that could be mixed and matched to support innovative uses. AWS is now building a ginormous set of Legos to support a new generation of applications.
The danger for incumbent vendors is that they will repeat the mistake of the railroads by believing they are in the infrastructure industry, not the application enablement industry. Many of them have looked to the size of the traditional infrastructure revenue stream and their position in the infrastructure landscape and dismissed AWS’s threat.
What incumbent vendors need to recognize is that the IT market battle is rapidly shifting from infrastructure to application enablement — and that market will ultimately prove to be far larger than the current market, just like all previous IT platform shifts. Missing this shift means becoming the next generation Penn Central.
Frankly, I’m surprised by the lack of urgency on the part of incumbent vendors. While they appear to have some awareness of the reasons for AWS’s popularity, they display a puzzling lack of intensity in terms of the pace of their response.
The evidence is plain to see. The largest vendors are experiencing dropping revenues or, at best, anemic growth — this in a time of obvious explosion in IT use. We are experiencing unprecedented growth and innovation in IT and, right now, the incumbents seem at a loss as to how to respond. One thing is clear: trying to fight this battle on the basis of infrastructure is a losing proposition.
AWS customers use it to help them deliver next-generation applications. Failing to understand this, and figuring out how to respond to customer demand for application-oriented functionality, will consign vendors to the scrapheap occupied by passenger trains and buggy whip manufacturers. This is a humbling prospect for yesterday’s IT giants, but as the saying goes, the first step to getting yourself out of a hole you’ve dug is to quit digging.
Named by Wired.com as one of the 10 most influential people in cloud computing, Bernard Golden serves as vice president of strategy for ActiveState Software, an independent provider of CloudFoundry. He is the author of four books on virtualization and cloud computing, his most recent book being Amazon Web Services for Dummies. Learn more about him at www.bernardgolden.com.