How to analyze and not analyze a behemoth:
Like many people, both inside and outside of the financial world, I have been watching Amazon (NASDAQ:AMZN) for a long time now. Back in the days of the tech bubble it was a favorite stock to trade on the short side. After all, how could a company that sold something as mundane as books, and sold them for less than they could be bought anywhere else, ever evolve into a profitable company? Over the years, I have watched, and been an Amazon customer, as the company increased the scope of its retail business and radically changed the way most people in this country think about online shopping. And yet I have wondered for years, at the expenses involved in building the company's franchise. It has not proven easy to really analyze Amazon's retail business because it seems to have so many ancillary and moving pieces as to come close to defying rational observation. I absolutely marvel at the complexity of the earnings models that many diligent and thoughtful analysts have put together. I recently saw once such analysis from CSFB that went on for no less than four pages of closely spaced numbers. I'm sure that the analyst in question took pains to make reasonable estimates for all of the numbers in all of the models but there is simply no one alive who can hope to be accurate in trying to estimate so many independent variables over a multi-year span.
The analyst at CSFB is hardly alone in his methodology and I suppose at some level institutional investors expect to see such methodology as a talisman to which they hold tight in order to justify their purchase of Amazon shares. Because the exercise is so thoroughly daunting, what happens is that many analysts and investors attempt to focus on quantitative metrics that they then trend going forward. That is always a dangerous thing to do in the tech space as so many black swans can and regularly do fly onto the playing field.
Some years ago, as most observers realize, Amazon got into the IT service business. It was and remains a brilliant decision. Web Services is really the driving factor in Amazon's future growth and its financial metrics are going to be the most salient in considering the future course of Amazon's blended valuation. Not quite yet, to be sure, but almost and in a few years, AWS will be the tail that wags this dog.
At the moment, AWS accounts for about 7% of total company sales although a far more significant share of operating income. AWS revenues are currently growing at about 70% year on year which is more or less unprecedented for an operation that is approaching a $10 billion/year run rate. These days, people are trying to determine if that kind of growth might continue and if so for how long and at what level of profitability. It is not an easy exercise because there really are no historical analogs that one can use. I have written about quite a few cloud software companies and even a cloud hardware company or two. Their growth rate at scale always slows down. I won't address the issue of profitability, which is quite controversial for some readers, but AWS already is a significant contributor to margin trends for Amazon as a whole. Just in the last year, AWS contributed 40% of the company's sector operating profits, up from 23% the year before. AWS margins last quarter continued what is seemingly an inexorable trend, rising to 29% from 25% the prior quarter and 17% the prior year. I think it is evident that absent the after-burner effects of AWS, the financial picture at Amazon would be cloudy at best and possibly quite stormy. If an investor is going to own AMZN shares, then he/she better be convinced regarding the longer term prospects for AWS.
In recent months, controversies have emerged as to the future course of AWS as numerous competitors have begun to vie for the No. 3 position in this space. Microsoft is a strong second and worth some further commentary. Google (NASDAQ:GOOG) (NASDAQ:GOOGL), IBM (NYSE:IBM), Oracle (NYSE:ORCL)and others are all trying to become meaningful competitors. Amazon has lost some high visibility customers as well including most recently Apple (NASDAQ:AAPL) which has defected to Google.
And Amazon shares have come under pressure of late. The shares, which enjoyed fantastic performance last year during which they went from $310 to a closing year price of $675, have retreated to just under $600. Part of that has been the overall downward trend of tech valuations, particularly those with companies that are highly valued in the eyes of many. In addition, Amazon's earnings release of Q4 results was seen by many as disappointing high expectations. Indeed, even though AWS revenues were ahead of forecasts, there were some who thought that the 69% growth rate in the quarter should be a cause for alarm. It was down after all from 82% the prior quarter and 78% growth the prior year. I will comment about current AWS trends below, but I confess to not remembering a time when 69% year-on-year growth was described as disappointing. The net of the earnings release which did not live up to prior optimistic expectations was that the shares lost almost 8% of their value the day after the earnings release and are still about 5% below the level that was attained just before the company reported its Q4 earnings.
In this article, I'm going to focus exclusively on AWS which really has little or no correlation with the company's retail business. I'm going to try to forecast some of the trends that are visible and to look at some of the recent negative conversation that has sprung up regarding the future for AWS. What I'm not going to do is to discuss the reasonable value for the shares as a whole. There are some articles I have seen on this site discussing their viewpoint that Amazon shares are overvalued - I am not going to attempt to refute them in any particular fashion. There is a species of thought that suggests that Amazon shares are "over-owned." I am not too sure what that means in practical terms. Amazon is 67% owned by institutions and 18% owned by insiders. Many, many companies these days have even greater institutional ownership percentages. That is about all I am going to say about the company's valuation as a whole.
I think that taking a deep dive into AWS is more than enough for a single article. I do believe that much of the earnings growth for Amazon as a whole will be coming from AWS. I think that a decent understanding of AWS will provide readers with a tool to make some reasonable guesses regarding the future valuation parameters for the company as a whole.
What is Amazon's real market share in Web-based services and how does it defend that share against a rising tide of competitors?
Amazon Web Services is a rather complex set of more than 50 different services these days so trying to figure out the total available market and Amazon's share is not a straightforward proposition. AWS will actually celebrate its 10th birthday this year. One can read lots of research from lots of different sources regarding the size of the market. One issue is that the different sources inevitably are not trying to measure the same thing. The lowest number that I have seen is that the TAM for AWS is $238 billion by 2020. I have made all of the necessary comments about the precision of such estimates earlier. What any reader needs to know is that the TAM is very large and it seems to be rising in terms of estimates over time. If you like $235 billion or $250 billion by 2020 it is all the same thing. Again, the low estimate I have seen for AWS by 2020 is that it will be a $20 billion run rate business. That implies a CAGR of just 20% falling to below 15% by 2020 but perhaps the analysts who wrote the piece were uncomfortable using a greater number. It seems almost inconceivable to me that if the TAM is really $235 billion that it would be possible for AWS to have less than a 10% share in 2020 when today it has a 27% share which appears to be rising.
I think a better estimate comes from Statista which uses data compiled by Wikibon. Basically, what AWS does is to provide users infrastructure in the cloud. This market is called Infrastructure-as-a service (IaaS) and Platform as a Service ((PaaS)). The study concluded that AWS had 27.2% of the combined market and the next five vendors combined had 37% and that everyone else had a 35% share. Gartner maintains in its study on the subject that Amazon offers as much computing capacity as the next 14 players in the market combined.
Gartner says that AWS is on track to be a $50 billion/business by 2020 and that makes a bit more sense to me. If the TAM is $238 billion, and AWS holds its current share, it would reach $64 billion in revenues. So, a $50 billion estimate allows for Amazon to lose some market share to start-ups and to the other large vendors in the market. Of course, $50 billion in revenue by 2020 would be a growth rate of 45% over that span. It would be a phenomenon unseen in American business literally since the days of the automobile or television.
I think it is probably important to understand why customers choose AWS and why they choose cloud services at all. I will expand on this a bit but simply put, AWS is simply cheaper, simpler and more reliable than any other way most users can get the same amount of computing. Cheaper, simpler and more reliable works for almost everyone. The current Gartner Magic Quadrant report shows AWS with one of the largest leads that I have ever seen in that series. The No. 2 in the report MSFT is said to have almost the same completeness of vision but far less executing competence.
When AWS started it was mainly for small developers who wanted to build "cool" applications but hadn't the resources necessary to fund the kind of compute capacity that was necessary to develop their dreams. So they came to AWS in droves because they literally had nowhere else to go. AWS still dominates that business segment but has now turned to the larger and potentially more lucrative business of supplying its services to the largest enterprises. Over the years, Amazon has basically chosen to invest everything it could into building larger and larger data centers. Some observers describe Amazon's strategy as the "virtuous cycle." Of course there are analysts who have written lengthy reports regarding the capital intensity of the AWS business. It is a somewhat capital intensive business but as it scales the intensity has diminished. Yes, capex for AWS is part of the factor that underlies some investor uneasiness about potential profitability but as long as the ingredients for the "virtuous cycle" remain intact, it is the appropriate strategy for the company. That is what AWS does and it seems likely to continue to try to make the maximum investments that it can for the foreseeable future. To the extent that readers or investors feel uncomfortable with such a strategy, then AMZN is not a stock to own. If on the other hand, readers and investors like "virtuous cycles" then AWS is one of the best exemplars of that strategy going.
Again, large data centers are really massively less expensive to build and to operate than smaller data centers per unit of compute power and can afford to use proprietary technology because their scale is just so great. The technical term for that is hyper-scale and everyone in the business aspires to be building hyper-scale data centers. Again, to restate the point, hyper-scale is lots cheaper per unit of compute power than anything else around. It is very hard to justify not taking advantage of the costs of hyper-scale and Amazon has done so in spades. Again, simply put, Amazon is able to offer compute power including storage and servers far more cheaply than its current competitors. It also is said to be more reliable, easier to use and has more available features.
The other advantage of scale is that it has allowed Amazon to steadily increase the set of features that it offers and to thus smother some of the potential competitors who simply are struggling to catch up with Amazon's capacity, let alone its functionality. Amazon says it added no fewer than 516 features to AWS last year. I have no idea what all these features are and even if I did I doubt that it would materially advance my evaluation of the forward momentum of AWS. But the point is that AWS market share lead means that it can amortize development costs over a significantly large base of revenues making more attractive and affordable for the company to add features.
As I mentioned I'm not a particular fan in drilling down too far in an analysis. But it is very hard to ignore what AWS is doing to the storage and the database business. Much of that $238 billion TAM is coming from the donations of "market share" from both storage and database vendors to AWS and the other participants in the field. One of the nice things about IAAS is that users can place a workload, literally any workload the service and see how it works. Users needn't lock themselves into some kind of minimum usage or to a multi-year plan. Because it is offered as a service, and not a hardware purchase, users can eliminate data centers and all of the ancillary costs and management hassles that operating data centers entails.
I don't mean to suggest that all workloads are going into IaaS networks but I do imagine that there will be very few users who will not bring some meaningful amount of their computing requirement to IaaS over the next few years. Even the most reluctant of users are going to be impelled to make changes in their workload distribution simply to keep costs competitive with those of other enterprises.
I think another statistic that I need to share is Gartner's projection of what it describes as Use-Case Graphics. Use case graphics really means how good the product is for application development. AWS gets a 4.81 score, Microsoft gets a 3.48 score and then come all the rest. Amazon has a 4.81 score for batch computing - that is where the user simply transfers his workload to an IaaS platform. MSFT has a 4.15 score, Google has a 3.72 score and then come all the rest. In what Gartner calls "cloud native applications" here are the comparative scores: AWS 4.84, Microsoft 3.99, Google 3.58 and then all the rest. I don't want readers to think that the other services that are offered are completely deficient. Most vendors do have what is called resilient infrastructure. All of the vendors use what is called VM clustering and have the ability to rapidly detect VM failure and restart it immediately on different hardware. Almost all of the competitors have a variety of plans for obtaining compute power and other resources that is chargeable for each VM at some hourly rate. AWS despite having the lowest costs per unit of compute power does not and need not offer the lowest prices. Why - well this is what Gartner has to say: "AWS appeals most strongly to customers who value thought leadership, cutting-edge capabilities, or a "safe" provider that has a well-proven service and is likely to continue to be the long-term market leader.
Think of it this way if you will. Cloud-based computing is massively less expensive than either traditional or hybrid cloud topologies because of scale economies. Amazon, with a huge lead in terms of compute power, has costs that are significantly less than its competition and because of the scale of Amazon it has developed features that no one has yet to match. In the south where I have spent much of my working life, the expression we would use is "case closed, next case."
Okay, enough already. I get it - Public Cloud is good, Amazon is better and it looks like things are going to stay that way for a while. But what is going on competitively? Any thoughts on what might happen to dethrone AWS?
Before launching into some specific comments regarding competitors, I think it is well to attempt to describe to readers one of the less-heralded advantages that Amazon has - less heralded I suspect because the advantage has nothing to do with Amazon really but it has to do with much of the competition in the segment. Again, simply put, AWS is not disrupting any legacy business or businesses at Amazon. Amazon doesn't sell database software or make storage boxes or sell application software. The company is just now putting its toe in the applications market with what it calls QuickSight. No prizes for guessing who that is going to compete with but if you need the answer here it is basically Qlik (NASDAQ:QLIK), Tableau (NYSE:DATA) and Microsoft as well as the legacy BI tools offered by the older stack vendors. It would be a far more difficult decision for legacy vendors to introduce something like QuickSight because of concerns regarding cannibalization. AWS has no such concerns.
AWS is already disrupting the database market and has become a leader in Gartner's magic quadrant for DBMS. When Oracle or IBM or Microsoft consider their public cloud business they have to consider the extent to which growth in the cloud will penalize their revenues coming from legacy offerings. It is not an enviable set of choices to be forced to make.
There were press reports yesterday to the effect that IBM is laying another several thousand people and that it is also buying a cloud consultant company called Bluewolf. How is that relevant in this discussion? Basically, IBM is trying to pull off a feat that is comparable in difficulty and complexity to the feat of taking one of those unicycles across a high wire and then trying to stand on its hands. Instead of simply trying to maximize revenues and profits in its new endeavors, it has to figure out how fast it can grow those businesses while trying to retain some level of revenues and profits from businesses that are in long-term secular decline. The Seeking Alpha story published on 3/31 says that IBM is to lay off 17% of its US tech sales force to make room for new "Open Source Positions." I challenge anyone, let alone IBM CEO Virginia Rometty, to pull that one off seamlessly. What it means in terms of this discussion is that legacy vendors are trying to compete with AWS with at least one hand tied behind their back. The competition really isn't fair and not hardly surprising that AWS continues to win a disproportionate share of the opportunities for which it competes.
I have mentioned that AWS has a lower cost structure than its competitors primarily because of its scale. What is perhaps underappreciated is that the lower costs of AWS also ensures that it can win any price war started by competitors. I believe that it has become a popular statistic to talk about Amazon's 49 price cuts since it entered the business in 2006. I don't propose to discuss all the different ways that users can buy cloud services and in any event, the price sheets for the vendors are not meant for potential large enterprise users. AWS has a service offering that is something like that of Priceline (NASDAQ:PCLN) in that it allows users to bid on spare capacity similar to a Dutch auction methodology. I think that there is little doubt that Amazon will continue price cuts and it will get more aggressive depending on price cuts from its competitors. Because Amazon has lower costs and more features, no one these days can hope to win a price war against AWS. In the last year or so, Oracle cut its prices in the areas in which it competes with AWS by no less than 90% according to analysts. And yet it is apparently still losing share points to AWS despite its triumphalist pronouncements on its last earnings call.
Microsoft was simply late to the market. It didn't introduce Azure until 2010. There is a reason for that. Customers who migrated to the cloud wouldn't need as many copies of Windows Server and SQL Server which are huge revenue streams for Microsoft and are used in largest companies' data centers. Microsoft has famously pivoted and it is surely the No. 2 competitor in this space more because of its installed base and customer relationships than its technology. It has been quite successful in using its existing enterprise agreements with current users as a lever to aid the adoption of Azure. And it has decided to support Open Source after years of trying to crush it. If there is any competitor that has some chance of catching AWS it would be MSFT. At this time, in the public cloud, the race is all in Seattle.
Google's Cloud Platform actually does have the scale to compete against AWS since it already runs Google's search engine, Gmail, YouTube and everything else that Google offers. And it has some pretty cool technology as well called Kubernetes management which it offers for free. So far, Google has simply been unable to get its go-to market messaging right and hasn't developed a specific reason for users to do business with it as opposed to AWS. Google gets almost all of its profits from selling ads so it has a huge cash flow with which to fund its cloud platform. It just doesn't quite seem to want to do everything that it takes to challenge AWS.
The two major legacy companies trying to enter the cloud derby are IBM and Oracle. Both companies are pushing something called hybrid cloud which is a reasonable solution for the largest of organizations that want to retain some compute power in-house. But over time, more and more companies are going to come to realize that it simply pays to outsource and that means going to the pure cloud providers. The cost differences are too large for hybrid solutions to persist indefinitely.
Forrester Research did one of its studies regarding the possible place for legacy players in the public cloud. Their conclusion was that "it's not quite clear what place there is for legacy players in the public cloud." When it comes to Oracle, the situation is hardly as it is represented on those risible earnings calls. Oracle can and does sell to its installed base. That it uses questionable sales practices by which I mean tying and threats of audits (many users have "free" or phantom database seats which is why some of the users wound up choosing Oracle in the first place) is well known. As I mentioned earlier, Oracle is big on discounts - real big. As many users might imagine, that is a two-edged sword in that the user community talks amongst itself and it hard to quarantine discounts of such levels.
But the problem is that its "cutting edge" users are leaving in droves, for CRM, for Workday and for database offerings that are pure cloud. Oracle makes some outrageous claims and some people even accept its goofy math, but the fact is that it has been and remains a share donor overall and that it just cannot reach the broader developer community with what it is selling these days.
There have been some industry rumors to the effect that Oracle, realizing just how down-market its current web offerings are, is investing heavily to build a new web services product that is supposed to come to market in 2017. It is said to be more reliable and to have higher performance than AWS and is to be focused on selling to banks, hospitals and other customers with stringent and security and compliance needs. So, in about a year, expect some further noise from Oracle about its new prize initiative. That is if the initiative remains on schedule and meets performance targets. This is software after all and those things are hardly a given. How might a speculative new product impact AWS if indeed it does come to market and replace Oracle's current cloud? I think in this instance the hypothetical is simply impossible to answer although I would bet the Amazon track record against the Oracle track record any day. But beyond that, I really don't know.
So, wrap it up here. What does the financial model look like for AWS and what does that potentially mean for Amazon? That is all we really need to know!
I hope that in writing these kinds of articles I have made provided a reasonable framework from which to construct AWS forward-looking financials. AWS may only be 7-8% of AMZN revenues but it is already 40% of the company's operating income. Without the margin over attainment of AWS in AMZN's last quarter, overall results, which in any event were seen as disappointing, would have been noticeably worse and the shares would have surely suffered more adverse consequences than has been the case.
I don't know how many readers are prepared to believe that AWS revenues will really reach $50 billion by 2020. I think I have presented an argument as to why they should, but forecasting that kind of outsize growth for such a large enterprise is unsettling to many readers - and frankly to this writer as well. I suppose there is something in us humans that wants to see forward projections regress toward the mean. I think it is also intellectually challenging to forecast something that has never happened in the past.
Many analysts have forecast revenue growth for AWS at just over 40% for the next couple of years and if 40% is a reasonable growth number to use that implies 2020 revenues of perhaps $35 billion to $40 billion. (Presumably revenue growth percentage will degrade over the years as has been the case in other IT segments.) Many estimates for the EBITDA margin by that time are north of 50%. That is almost double current margins, but I think that it is quite possible that such margin assumptions will prove to be conservative.
Many people note the price competition in this space about which I have commented. What the people who comment about the price deterioration seem to have missed is that costs for AWS are contracting more rapidly than pricing. The "raw material" for what this company sells continues to plunge in price. Without a lengthy discussion regarding the costs of servers and storage and switches, I think most people understand just how significantly the cost of those components is dropping. And on the pricing side, this company is selling some value added services and that trend is sure to continue. I commented about the initiative that Amazon has launched in data visualization. Forrester has positioned AWS in the leader quadrant for Enterprise data warehousing, a far "pricier" subset of the basic cloud service offering. And Gartner has positioned AWS in the leader quadrant of DBMS (Database) where the competition is actually Oracle whose margins in that segment have been extraordinary and Microsoft. In February, an organization called 451 Research declared that AWS is a leader in the Big Data segment, another significant value-added and highly differentiated offering compared to those of current competitors.
Noticeable as well are those companies who have withdrawn from competing for public cloud business. In the last several months HPE (NYSE:HPE), Verizon (NYSE:VZ) and Rackspace (NYSE:RAX) have given up the ghost and dropped out of the market. I would suggest that the Web services industry will consolidate significantly over the next few years, potentially putting both a floor under prices and a floor under the AWS market share as well. I would think that the odds favor greater growth in margins beyond 50% EBITA but again we are totally within uncharted waters here.
Oracle has an EBITDA multiple of 10.6 and CRM has an EBITDA multiple of more than 80. What is the right EBITDA valuation to use for this company? Obviously, part of that answer is how fast AWS might be growing at the terminal year in the forecast. Oracle isn't growing at all and hasn't grown in several years. And CRM's growth is supposed to slow to 20% by 2017 although there would be some, including this writer, who would argue that growth is likely to be greater at that point. I think that in the case of AWS, an EBITDA multiple of about 15X seems reasonable given all the unknowable issues there are regarding 2020. That suggests that the valuation for the AWS business ought to be something on the order of $300 billion at that point.
At this point, most analysts would discount that sum back to present value but the discount rate for the exercise is almost impossible to choose. Discount rates are usually a function of risk, which I think is pretty low coupled with some central tendency interest rate which has to do with overall interest rates. I think the product of such an exercise probably produces an AWS valuation of at least $200 billion using what I think are quite conservative assumptions. Readers might want to quibble with certain points in the analysis, but at the end of the day the space and the AWS leadership are really not in that much question.
Amazon has a current market capitalization of $279 billion. Is there anyone who thinks that the other segments of Amazon aren't worth more than $79 billion? Most of the estimates that I have seen suggest that the rest of Amazon is poised to achieve more than $115 billion in total revenues. I think that has simply got to be worth more than $79 billion-probably several times more.
Without qualification and using what the writer believes to be conservative estimates based on the analysis of the vast preponderance of industry analysts, Amazon shares are quite undervalued. If ail the assumptions regarding AWS growth bear out, it seems quite likely to me that investor sentiment will turn positive again on this name.
Some final thoughts:
I have tried to present an analysis for the prospects of AWS in some level of detail. As a business, AWS is unprecedented. There has never been a growth ramp at scale that even approaches what is happening at this business. And the ramp is predicated on just a few more or less undeniable facts. The AWS solution is far cheaper than alternatives, it is more reliable and there is significantly less hassle for enterprises in outsourcing as much IT infrastructure as they can. There are very few enterprises that really want to run data centers - it is no one's core competency outside of the companies already in the IaaS market.
Again, the market share that AWS has achieved gives it the funding to keep pushing down costs through scale and to add more and more features and value added components to its business. I think that is more probable that AWS will actually increase its market share from the current 27% level over time rather than the other way around. But the numbers get so large as to be totally off-putting. In trying to estimate future revenues and EBITDA margins, I chose assumptions that are clearly less than rather than greater than either current trends or any kind of qualitative analysis. And the product is still $200 billion.
There are many, I know, who fault this management for not paying enough attention to margins or returning capital to shareholders. There are others who delight in finding blemishes on the body of this behemoth. I really don't blame people for looking carefully at what this company is doing. It is fun and intellectually gratifying to discover chinks in the armor of icons. But AWS is a very large company. It obviously is going to be doing something poorly or sub-optimally. But what those things might be in AWS is not readily apparent to me at this writing.
Many years ago, I lived in Oklahoma and occasionally in the Spring, I would go out on a hill near my home and watch a tornado settle on the plain below. It was really an awesome sight, but scary at the same time. The sound is like nothing anyone can imagine and the sky is lit up by lightning flashes that reflect from the undersides of the clouds. Of course, some tornadoes disrupt the communities through which they pass. Watching the sight, awesome as it can be, is simply an acknowledgment of God's unknowable ways.
To my mind, AWS is like that. It is making loads of noise, it presents an awesome spectacle and it is ripping up the competition as it goes. The closest thing to - a force of nature you will ever see in the IT world.
Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.
I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.