It is time for Amazon Web Services to get out of Amazon

Andy Jassy, AmazonAmazonAndy Jassy

A recent interviewwith Amazon Web Services’ chieftain Andy Jassy repeatedly touts the “trillion dollar opportunity” associated with the inexorable transition of enterprise IT to the cloud. Perhaps because his interlocutor was so busy inserting himself into the interview, Jassy isn’t actually quoted uttering the T-word (although he does manage to land “TAM”). But he certainly isn’t refuting the notion there are thirteen orders of magnitude associated with the Enterprise Cloud Jackpot (and I use Jackpot in a thoroughlynon-Gibsonianway, though legacy vendors may see it apocalyptically). As the cloud computing leader today, AWS is in the pole position for this tectonic shift. The question is whether being part of Amazon going forward is on balance more help or hindrance to AWS winning their fair share of the Jackpot.

Amazon obviously was able to conceive, nurture and build a heretical idea into AWS today. It is inconceivable any traditional IT vendor could have done this and hard to imagine other candidates with comparable wherewithal and “willingness to be misunderstood”. But we’re past the point of being misunderstood. Pretty much everyone now accepts that vast chunks of IT are shifting to the cloud (with the notable exception, admittedly, of legions of enterprise IT people clinging to their servers in a desperate bid to preserve the status quo). Even the biggest technology dinosaurs now recognise that the incoming cloud meteor is a potential extinction event.

AWS continues to innovate and execute well. Their near and medium-term success is not in doubt. But being part of Amazon raises questions about their ability to fully capture the big prize. Can they achieve generational dominance a la IBM or Microsoft in their primes? Will they be one major player among several? Or does their early lead whither away? Being part of Amazon brings benefits, but also significant baggage, specifically difficulty concentrating, having sufficient cash to compete and adverse cultural factors.


Big as it is in its own right, AWS is a relatively small part of Amazon and gets lumped into their glamorous “Other” segment, which did $US5.6B in 2014 (about 6% of Amazon’s overall $US89B in revenue). Notably, “Other” was the fastest growing segment at 42% vs. 20% for the company overall. The Amazon-branded credit card gets cited as another occupant of “Other”, but our grasp of the AWS business remains tenuous.

Beyond the trillion dollar Enterprise Cloud Jackpot, Amazon’s core e-commerce business still has tremendous upside. E-commerce in the US last year was over $US300 billion and but only 6.5% of overall retail sales. Amazon takes an expansive view of the ecommerce business, investing in drone and local delivery, a broad line of consumer hardware for direct distribution of digital goods, a robot army and private label brands for everything from diapers to TV shows.

In contrast to Apple who says “no” by default, when confronted by new opportunities Amazon seemingly leaps up on the table and screams “Yes! Yes! Yes!” Their bonfire of initiatives has resulted in some notable and expensive misfires of late, including the disappearing Kindle Fire, the disastrous Fire Phone and the under-heralded Fire Diaper blowout And the jury is still out on Fire TV, the not-exactly-setting-the-world-on-fireEcho Fire, The Man on Fire in the High Castle, WorkMail Fire, and “hair on fire” one hour delivery amongst other efforts. And, for completeness, this is an AWS Fire.

It can be argued the company has poor impulse control, is fighting too many battles on too many fronts and must allocate cash, talent and attention across a very wide range. Amazon’s historical “willingness to be misunderstood” and success in defying past criticism (to which I hereby add my name to a long list) don’t do anything to suggest new restraint is imminent. This scattershot approach may make sense before product-market fit, but AWS is beyond that point. Many of these other activities are a distraction given AWS is chasing a plausibly trillion dollar market.


Cloud computing is not for the light of wallet, requiring many billions of dollars to build cavernous data centres scattered around the planet, stuffed with millions of servers. GigaOmbreaks down last year’s capex spending by the big players:

Now these numbers aren’t pure apples to apples comparisons. Amazon’s capex also goes to erect even more cavernous e-commerce distribution centres with “more than 15,000 robots in 10 fulfillment centres across the U.S.” Google and Microsoft both spend big on search infrastructure, and Google’s numbers include YouTube, flotillas of automotive, sea-going, aerial andspace-faring vehicles, and based on the sheer magnitude, maybe the odd space elevator.

For comparison (and in lieu of a much-delayed “So You Want to be a Cloud Wanna-be” post with etiquette and strategy tips for cloud also-rans), here are similar period capex numbers for some of the cloud wanna-bes and their growth (or lack thereof) over the preceding year as they try to play catch-up:


Growth over 2013
















(These are the budgets from which bonsai datacenters are built).

Meanwhile, returning from Lilliput, Amazon has the shortest stack at the grown-up table, with $US17.4 billion in liquid assets to Google’s $US64.4 billion and Microsoft’s $US90.2 billion. Google in particular intends to push Amazon to the wall financially, though it was Bill Gates who said, “Never get into a price war with someone who has more money than you.” Amazon has notably lost its prior enthusiasm for price cuts and obliquely grouses about networking costs where Google and Microsoft have a huge advantage in terms of facilities.

It has always been a bit of a mystery how Amazon could serve more cloud computing customers, spend less than Google or Microsoft and support aggressive build-outs of both datacenters and distribution centres. The assumption was search required significantly more infrastructure and/or Amazon was just somehow more efficient. But we’ve recently learned that Amazon has used capital leases to keep billions in capital spending off their cash flow statement. So their capital spending is closer to double what has been reported. This means their free cash flow, the metric they have relentlessly told investors is the key lens onto the company, is actually negative and declining in recent years (remember this: we’ll revisit it below when we talk about employee recruiting and retention).

It is an easy quip as a low margin retailer to say “your margin is my opportunity” but you still have to pay for your datacenters. Every dollar spent on half-baked consumer hardware (or worse,doubling down on a failed consumer hardware brand) or random TV shows, or giant distribution centres and bright yellow robots to support the core e-commerce business, is a dollar that isn’t going to AWS. And Amazon must continue to invest significantly in its core e-commerce business, whether defined narrowly or broadly.


It is an open secret, at least in Seattle and amongst top tier technology firms, that not many people enjoy working at Amazon. Many companies have come north to open offices (e.g. DropBox, Facebook, Palantir, Salesforce, Twitter) and usually say they are there to lure talent from Microsoft. Yet with the notable exception of recent SoCal immigrants Oculus andSpaceX who are on Microsoft’s side of the lake, most choose to locate their offices a stone’s throw from Amazon’s campus.

Amazon relies heavily on its stock for both recruiting and retention. They pay below market salaries, offer signing bonuses that pay out over the first two years and after that rely on stock to both carry the compensation load and overcome the drawbacks of the work environment. Independent of stock performance, you see a lot of Amazon resumes on the street as soon as the signing bonuses are paid out.

Amazon’s stock price obviously reflects perceptions of the company broadly. It has been run as a profitless machine yet been richly valued for most of its history. Scepticism has emerged over the last year or two about whether the company will ever get out of “investment mode” and focus on financial returns. Amazon turned in a strong Q4 including an unexpected profit (they have conditioned Wall Street well…) which popped the stock by over 25%, but it still trades below its all-time high even as the rest of the market hits record highs.

Amazon has guided investors to focus on its free cash flow as opposed to profits. Yet they felt the need to more explicitly call out the significant use of capital leases in their most recent earning call, which had heretofore been buried and largely ignored amongst the footnotes. The free cash flow picture looks very different, in both magnitude and trend, when these leases are considered:

And note this debt is being incurred today at unprecedentedly low interest rates, a phenomenon that is unlikely to continue much longer.

AWS is growing fast and needs to continue to hire rapidly as it chases the Jackpot. Dependence on Amazon’s stock, and the company’s ability to sell Wall Street on anything-but-profit metrics, adds risk to their ability to recruit and retain. AWS also needs to build a culture that can meet enterprise customer expectations and that culture surely looks different than today’s Amazon.

For example, Amazon’s secrecy fetish and chronic inability to put numbers on the y-axis of charts is inconsistent with need to provide long-term roadmaps for customers contemplating taking enormous dependencies on AWS. The secretive culture alienates a whole range of useful hires for AWS. Further, Amazon the retailer gets confused about what it means to be a platform company. AWS has generally done a decent job walking this fine line (and platforms and their ecosystems are never static, which is a topic for another day) but a deft touch in nurturing the ecosystem is critical to AWS’s future success.

Set the Cloud Free

On balance, I think AWS would be better positioned to realise the opportunities in front of it as an independent company, away from the vanities and vicissitudes of Amazon.

A full spinout of AWS from Amazon or even a partial IPO a la VMware would establish a focused entity that can single-mindedly pursue the Jackpot, with its own culture and compensation model. Amazon can stake the new company and the transparency associated with being a pure play chasing a $US1 trillion TAM should give it the opportunity to raise all the money it needs to compete, especially while growth is still in the mid double digits. More distance from Amazon would also help AWS land more customers. Many large retailers consider AWS verboten and as Amazon’s sprawl continues, that competitive dynamic could impact AWS’s ability to sell to media, consumer products and other industries.

A spinout of AWS may already be in progress. Amazonannounced it will start disclosing AWS numbers this year, which could be a first step. These numbers will be scrutinized every which way and will be fascinating to see. We could learn that it is AWS’s capex specifically that is being capitalised and thus intended to be portable. Rumours also suggest AWS gets far less attention from the chief product designer and octopus taster at Amazon.

Will AWS still be part of Amazon in two years?

This article originally appeared on Platformonomics. Copyright 2015. Charles Fitzgerald is a Seattle area angel investor and cloud watcher. Read more posts from Charles at Platformonomics. Follow Charles on Twitter.

Disclosure: Jeff Bezos is an investor in Business Insider through his personal investment company Bezos Expeditions.

NOW WATCH: Here’s the unlikely story of how auto-tune was created

NOW WATCH: Tech Insider videos

Want to read a more in-depth view on the trends influencing Australian business and the global economy? BI / Research is designed to help executives and industry leaders understand the major challenges and opportunities for industry, technology, strategy and the economy in the future. Sign up for free at