The Channel logo


By | Gavin Clarke 28th December 2015 14:02

2015 wasn't about AWS. It was about everybody getting ready to try to beat AWS

Will the end of cheap money cast a cloud over 2016?

2015 Year Review One star eclipsed all others in the enterprise in 2015: Amazon. Or, rather, its cloud division, AWS.

It somehow became a given that AWS leads in Infrastructure as a Service (Iaas), with the one-time ebook shop now a major IT infrastructure provider.

That ebook flogger has become an IT infrastructure provider for the likes of Ocado, Norwich Union, News UK, British Gas and the Co-Op Bank in the UK.

AWS’s services have gone far beyond humble compute and storage, and it currently has fingers in the pies of data warehousing, disaster recovery, content delivery networks and more.

The baton of fastest-growing AWS service passed from Redshift, AWS’s data warehouse as a service, to Amazon's relational database engine, launched in July.

Amazon claimed a 95 per cent year-on-year increase in virtual machine (EC2) instances and 120 per cent growth in data transfers from its storage service (S3).

Things have gotten so big, AWS announced its first UK region: a substantial commitment that’ll require several physical data centres.

But what did that mean where it matters, on money?

Drilling into the financials

After much excitement, mainly on Wall St and among tech’s media cheerleaders, Amazon began reporting how much money the AWS unit actually made.

Until this year, AWS was hidden in the North American “other” column of Amazon’s spreadsheet.

The inaugural figures, in April, saw net sales of $1.5bn versus $1.05bn in 2014. Two quarters later that number was $2bn versus $1.1bn – a change of 42 and 181 per cent respectively.

Profit more or less doubled between quarters – from $265m to $521m, up 96 per cent.

How did that stack up to the competition – the giant software providers lumbering into cloud after dominating the enterprise for decades.

From a fiscal perspective, comparisons are almost impossible. Everybody claimed they were a leader – that they were the industry’s fastest-growing cloud company.

But all hid behind the great reporting evil of run rate – to claim a size of future business extrapolated from today’s figures.

Run rate is about as financially and mathematically sound as the numbers behind a subprime loan. Even Microsoft’s genial ex-chief executive Steve Ballmer took the firm to task for hiding behind run rate rather than saying what Microsoft’s cloud was actually selling.

He spoke after Microsoft cooked up something called Intelligent Cloud as a financial reporting category, comprising Azure Windows server and enterprise mobility. This saw eight per cent growth $5.9bn in Microsoft’s most recent quarter, the firm said.

Microsoft Azure revenue and compute use had “more than doubled” year-over-year along with in Office 365 its online productivity and collaboration suite, it said.

SAP and Oracle, giants of the enterprise, also reported massive growth in their respective cloud services.

SAP in October reckoned on 116 per cent growth to €600m. In its second quarter, reported December, Oracle claimed 26 per cent growth.

But, unlike SAP, Oracle’s core business was hit hardest by the slowdown in spending in on-prem business software.

Sales of new software and maintenance in SAP grew 11 per cent to €3.5bn ($3.9bn) in the quarter reported in October; Oracle saw its on-prem business – its core engine - consistently fall quarter after quarter. In December Oracle reported another quarterly decline, down seven per cent to $6.36bn.

Faring the worst of all, however, was the company whose products it was once impossible to get fired for buying. IBM, the one-time hardware, software and services giant, had missed 14 quarters by October, when it reported a 14 per cent drop in net income for its third quarter to $3bn.

Like others, IBM also mixed its cloud amongst “other” stuff, making unambiguous comparisons impossible: IBM’s cloud has been added with analytics and engagement to comprise a category it called “strategic imperatives revenue”.

This was up 17 per cent year on year by October, while sales in the company’s familiar bread-and-butter services, software and hardware, all fell.

Out with the old, in with the new

As in years before, AWS set the pace on features and development. AWS claimed 500 new features and services for AWS up to 6 October. That compared with 516 “significant” features for the whole of 2014 and 280 the year before.

The competition was left racing to catch up - rolling out their own, new capabilities or, in IBM’s case, acquiring companies to add features to their cloud.

Such was the pace at Microsoft that Gartner’s distinguished analyst Lydia Leong – the analyst’s go-to voice on public cloud – reckoned that Microsoft had become the firm number two behind AWS. To help cement that in the UK, Microsoft announced plans for a brand-new Azure data centre – its first in the country.

But, featurewise, Leong said, Microsoft’s Azure was no AWS. Rather, it was “good enough” – often lacking the completeness and polish of Amazon’s service. In terms of growth, she reckoned Microsoft was able to leverage its existing, on-prem relationship with customers to get a leg up into the cloud.

Oracle also threw out a brace of features around its annual OOW conference, while (the-then) Hewlett-Packard updated its Helion cloud with support for Docker containers.

HP, however, was hobbled by larger concerns.

The firm – which soon after undertook a historical split to become Hewlett-Packard Inc and Hewlett-Packard Enterprise – reversed gears massively on cloud.

First, HP said it would kill its Helion public cloud service in January 2016 – five years after unveiling ambitious plans for an all things-to-all-men hosted service.

Next, HP backtracked on its commitment to the OpenStack open source cloud – the basis of Helion – by becoming a premier partner of Microsoft’s proprietary Azure. HP reckoned it would offer Helion on its hardware according to customer “demand”.

The implications for OpenStack’s future are now unclear: HP had become the OpenStack Project’s single biggest committer, having hurled engineers at the project to polish its rough edges and make it friendly to enterprises.

HP’s move relegated OpenStack even more to the status of also-ran cloud platform. Once seen as the open alternative to AWS proprietary private cloud, OpenStack was fast becoming a thing for regional cloud providers and something used behind the firewall.

Now, who have we missed here? Ah, yes: Google – renamed Alphabet.

Back in 2008, the internet’s number-one ad slinger was close behind Amazon’s AWS with Google App Engine, which allows you to host apps on Google’s platform. Seven years on, Google is nowhere in terms of being a provider of public cloud services for enterprises.

In fact, conversations The Reg has had with customers revealed a deep-seated aversion to Google services, with CIOs telling us Google simply lacked the features and support they’d expect as a potential provider of their IT infrastructure.

And yet cloud is an important thing to Google, with the company’s senior vice president of technical infrastructure Urs Hölzle reportedly telling the Structure Conference Google’s cloud revenue would surpass its ads revenue by 2020. For a little perspective, Google’s cloud revenue is still listed under the “other revenues” section of Google’s balance sheet – with some estimating cloud to be worth $400m this year. By contrast, Google made $16.79bn from ads in its most recent quarter.

What to do? Appoint the former head of a firm that became essentiql to the enterprise, that’s what: in this case, ex-VMware chief executive Diane Greene.

Greene, a Google board member, made her name leading VMware though a period of expansion turning it into one of tech’s fastest growing companies and selling to EMC in 2003 for $635m; she was ousted by VMware owner EMC in 2008.

Greene in November took over the running of Google’s cloud with a remit to bring the ads-flinger’s cloud from far, far behind against AWS and Microsoft.

Larry Page, Google’s co-founder and chief executive of Alphabet, called Greene’s new post a “huge new responsibility”.

Private cloud

So much for public cloud, what of private? That was dominated by VMware, courtesy of its hypervisor. VMware retained its crown but, according to Gartner, Microsoft had closed the functionality gap with Hyper-V and was challenging VMware in SMBs.

As for public cloud, well, that remained AWS’s thing. In a further blow to VMware, HP killed its EVO:RAIL hyper converged love-child and then partnered big with Microsoft on Azure.

We're the fastest-growing... No, we're the fastest-growing... No, we're etc etc

What became apparent this year was that the field of challengers could not best AWS on features – not with so many racing simply to get basic services up and running.

And yet, they claimed growth. How was that possible? That’s because the real action of 2015 in the kinds of deals IT firms were willing to make simply to get customers signed up to their cloud service in some way, shape, or form.

Enterprise giants, The Reg discovered, were offering massive sweeteners to win customers – or, to at least prevent existing accounts becoming AWS customers.

Oracle, The Reg learned, was giving sales staff bonuses of seven times salary to flog their firm’s infrastructure, platform and software as a service.

InfoWorld later reported that Oracle gave away its cloud in 90 per cent of deals.

Oracle wasn’t alone: Microsoft was giving its customers 15 per cent off of planned enterprise software deals and giving away millions of dollars' worth of the Office 365 suite – even if they didn’t plan on using it.

Others gave up in the face of what seemed like the inevitable dominance of AWS. With Amazon moving deeper into online storage, analytics and warehousing, EMC fell into the arms of PC maker Dell in a record-setting $67bn deal.

Dell’s rival HP didn’t only kill its public cloud, but split and re-grouped into separate camps – HP Inc, around PCs, and Hewlett-Packard Enterprise, around servers, storage and cloud. It was like the splitting of the Roman Empire: which of the HP empires would survive longest?

So, for another year, Amazon goes out on top, thanks in part to its own achievements and in part to the weakness of and falling away of others.

AWS’s position was powered by growth and by rollout of new features.

That rollout is a virtuous circle of what some might call lock-in: the more features that are added to AWS, the deeper customers go into the AWS infrastructure and the harder it is to migrate.

For the majority of the AWS competition, the race became not to beat AWS - not even to come second, as Microsoft seems there according to Gartner.

It became a competition to determine third place in 2016 and beyond.

But that’s not to say AWS is unassailable. Sure, AWS dominates now, but others have dominated technology before and they have long since faded and disappeared.

What are the risks to AWS?

Well, it built new features – rolling out hundreds of advanced capabilities while everybody else raced to get their foundations in place.

AWS hasn’t stood still on price, either – cutting prices more than 50 times in different areas to keep up with Microsoft and Google.

When it comes to the playbook of how not to become an over-priced owner of a legacy technology empire ripe for disruption, AWS made the right moves.

Can anything stop AWS in 2016?

Possibly outside forces. In this case, the Federal Reserve.

As has become standard practice among those reporting, the tech flavour of the month is fêted with hagiographic coverage. AWS in 2015 could do no wrong.

The reporting on AWS focused on customer and revenue growth and operating profit – not net profit, or profit after costs, or costs such as interest payments.

Cheap money fuelled so much of tech’s success stories in 2015 in the startup scene – firms like Uber.

But cheap money is helping others, too.

The analyst firm Canalys highlighted what it believes is the unreported secret – and great weakness – of AWS: capex, which Canalys reckoned comes in at $3.5bn annually.

That figure is manageable if interest rates remain low, enabling cheap borrowing. US interest rates have been near zero per cent since 2008, the dawn of the financial crisis. A one per cent rate on $3.5bn means an interest rate of around $140m for AWS.

But a rise will inevitably mean increased costs for borrowers – eating into margins. A rate of, say, five per cent, would push that initial $140m figure up to around $700m for AWS. Translated, what does that mean?

Greater cost controls, profits eroded, restricted ability to invest in things like new data centre builds, less freedom to cut prices.

And so it was the US Federal Reserve ended months of speculation in December by putting up interest rates by 0.25 per cent. That mean US banks can now charge between 0.25 and 0.50 per cent to borrow money.

“When interest rates go up, that’s when we expect this will change”, Canalys’ Steve Brazier said: “Capital will become more expensive, and the cloud will become more expensive.”

That’s the problem with stars. They must all run out of gas some day. ®

comment icon Read 17 comments on this article or post a comment alert Send corrections


Frank Jennings

What do you do? Use manual typwriters or live in a Scottish croft? Our man advises
A rusty petrol pump at an abandoned gas station. Pic by Silvia B. Jakiello via shutterstock

Trevor Pott

Among other things, Active Directory needs an overhaul
Baby looks taken aback/shocked/affronted. Photo by Shutterstock

Kat Hall

Plans for 2 million FTTP connections in next four years 'not enough'
Microsoft CEO Satya Nadella


League of gentlemen poster - Tubbs and Edward at the local shop. Copyright BBC
One reselling man tells his tale of woe