This is the second in my series of essays on Amazon (NASDAQ:AMZN). In my previous article, I set out a valuation thesis for Amazon based on seven key numbers. In this and the next few pieces, I will discuss the reasoning behind the estimates of each of these seven numbers.
Start with Amazon's net sales. This is by some distance the most "amazing" Amazon statistic. As the chart below shows, net sales have grown at an average rate of nearly 31% per year over the past 11 years.
As revenues swell however, it becomes ever more difficult to maintain percentage growth rates. Since Amazon's sales crossed $50bn (in 2011-12), the rate of growth has slowed markedly, falling from the high thirties to the low twenties, even as Amazon has added about $13bn of additional revenues in each of the past three years.
To get a better picture for why this has happened, let's examine Amazon's sales mix. The chart below shows how Amazon's sales mix has shifted from being dominated by media in the early years to electronics and general merchandise in more recent years.
This changing mix explains a great deal of Amazon's slowing sales growth. The next chart shows the segment-specific growth rates.
Sales growth in media has slowed dramatically, from 30% in 2003 to 10% by 2013. Sales growth of electronics and general merchandise initially more than made up for this, but since 2010, even this category is growing at a much slower pace, falling from 66% in 2010 to 26% in 2013. Meanwhile, Amazon's non-retail services segment (which principally represents advertising and the cloud computing business) has picked up pace, accelerating from a growth rate of 20% in 2009 to near 60% by 2013. However, because media sales are six times, and EGM sales twelve times, larger than the non-retai lservices segment, the overall sales growth has declined over time to just over 20% per year.
To estimate Amazon's net sales growth rate in the future, we have to make assumptions about future growth in each of these business segments.
Media: Amazon has tried very hard to accelerate sales of media by improving the offering through its Prime program (through which members get unlimited streaming of movies on Prime Video and borrowing rights on a wide selection of Kindle titles), through sales of e-readers, tablets and smartphones, and promotions via its Goodreads website. The chart above reflects the fruits of this effort. Amazon has not been able to reverse the secular decline in media sales growth. In my view, it is therefore reasonable to assume that media sales growth will continue to fall in line with the historical trend, from about 10% in 2014 down to about 7% by 2018, and continue to fall thereafter to flatten out by 2023. The first two quarters of 2014 are in line with this trend, with media sales growing at about 8-9% over the previous year.
EGM: Forrester Research recently estimated that US online retail sales will growth at an annual rate of 9.5% per year from 2013 to 2018. Amazon's brand, selection, low pricing and logistics capability position it well to continue increasing market share in this segment. Its flagship Prime membership program - through which members get unlimited free shipping in exchange for a fee of $99 per year (£79 in the UK) - is a major part of this effort. I think a reasonable (if somewhat optimistic) assumption is that Amazon's EGM sales are likely to increase at double the growth rate of the US online retail market (i.e at an average of about 18-19% per annum over the entire period), but probably with higher growth rates in 2014 and 2015 (say 25-27%) falling gradually to about 13% by 2018. Thereafter, the growth rate of Amazon's EGM segment should gradually fall to about 3% by 2023 as Amazon reaches a degree of market saturation in the US. International sales could still drive growth in this segment in the longer term, but the trends are not encouraging - year on year international EGM growth has fallen sharply from 55% in 2011 to 23% in 2013 (reflecting weaker demand in Europe, but also increasing competition in Asia).
Non-Retail Services: Advertising and cloud computing (aka Amazon Web Services) are the two main services. Frustratingly, Amazon does not provide a breakdown of the segment, so investors have to try and guess for themselves. Pacific Crest Securities recently estimated that revenues from Amazon Web Services were $1.9bn in 2012, $3.1bn in 2013 and will reach $5bn in 2014, growing at around 60% per annum. Assuming that is a correct estimate, this would suggest advertising income in 2014 will be just over $1bn (and growing at about 30% year on year). Pacific Crest then goes on to forecast AWS revenues of $40bn by 2020, implying a growth rate of around 40% per annum in cloud computing. I think this is overly-optimistic - for instance, other previous estimates have estimated 2020 revenues at only $20bn.
I think it is more reasonable (though nevertheless still optimistic) to assume that Amazon Web Services could grow to about $30bn by 2020 (growing at about 35% per year), while advertising continues to grow strongly at 25-26% a year to reach about $4bn by 2020. Thereafter, growth should start to moderate in both cases, and overall services revenue growth will decline rapidly down to about 5% per year beyond 2023 (in line with the rest of the business). The performance of the non-retail services segment in the first two quarters of 2014 is broadly in line with this trend, having grown 27% over the same period in 2013.
These segmental growth assumptions produce the following forecast for net sales (sales growth rate is on the left axis, net sales by value are on the right axis):
This is the reasoning behind my thesis that Amazon's net sales growth rates will progressively fall, as set out in the first essay in this series. In the next article, I will turn to Amazon's EBITDA margins.
Disclosure: The author is short AMZN. The author wrote this article themselves, and it expresses their own opinions. The author is not receiving compensation for it (other than from Seeking Alpha). The author has no business relationship with any company whose stock is mentioned in this article.