The AWS business is apparently a roughly $6bn-a-year revenue business. That’s all external revenue from customers, not charges from other Amazon business units, according to the notes to Amazon’s 10-Q. The costs story is far murkier, as we discussed previously, so we can guess that the margins are less than 17%, but we really don’t know how much less.
All the other major public cloud companies – Google, Azure, Softlayer – are inflating their revenue figures in one way or another, by including various things that are arguably cloud or not cloud, depending on who you talk to. Mind you, there doesn’t appear to be quite as much scrutiny on AWS as there is on the others about this practice.
What Even Is Cloud?
I went to an event held by local integrator-cum-hosting-company Interactive last week, which was all about reinforcing their cloud credentials with their customers. I was the only media in the room, and everyone else seemed to be either Interactive, a vendor partner, or a customer. No prospects, so this wasn’t a “close the sale” event, but a “you’ve made the right choice” warm and fuzzies event.
There was a panel where various customers and partners expounded on how much they love Interactive, but during Q&A, I asked if what they were talking about what not simply a managed service style outsourcing arrangement, not ‘cloud’ per se.
What fascinated me was the divergence in opinion about just what ‘cloud’ actually is.
It’s Not About Price
AWS is definitely the leader in public cloud, largely because they got there first. Microsoft has done an incredible job in a very short period of time with Azure, and it really is a pretty compelling alternative, and for a bunch of reasons I’ll go into in a moment. Google just doesn’t seem to know how to market itself, so its version of cloud is lagging. IBM’s SoftLayer is an also-ran suffering from being owned by IBM. It’s the right move for IBM to be in cloud, but they missed the boat a bit, and, unlike Microsoft, just can’t get their act together enough to make it all happen fast enough.
So why do people choose AWS? It has very little to do with the price, or rather, it’s much more complex than price. And it’s why Microsoft’s Azure is the biggest threat to AWS, in my opinion.
AWS is not cheaper than on-site gear when you’re just talking about infrastructure. The chief advantage of AWS is speed.
Time is Money
AWS got its toe-hold with startups. It meant you could get started very quickly with something production ready, and with very little capital outlay. The now famous adage that “cloud makes it cheap to fail, but expensive to succeed” is actually about how quickly you can stop spending money if something doesn’t work. When you buy physical kit, you can’t give it back to the vendor if you decide you don’t want it in three months because your project didn’t work out. That’s really attractive for a capital poor startup, but it’s also the compelling reason for corporate development. The downside risk of doing a project is lower, and if things go well, you can more closely match the infrastructure spend with the growth of the application: you just rent more kit.
Contrast this with the traditional corporate IT approach of spending months doing a solution architecture. IT finance haven’t worked out accruals, so if you need ports (or storage, or servers, or databases, or VMs) but the switches/storage/cluster is full, guess whose project buys the next chunk of infrastructure? And that takes procurement, so now you have to wait for the gear to arrive, get installed, configured, and then you can start dev. Maybe.
Apart from the costs of the gear not aligning with the project business case (I only need 3 VMs! Why do I need to buy a new server and 2 switches?!) it takes ages to do. If your project is running, that means you’re paying for all the solution architects, the DBAs, the enterprise architects, security consult time, and for all the developers who are sitting around because you need to go through a third review because some EA didn’t like the colour of your Visio icons.
That is why business units hate internal IT and why, by comparison, the arms-length-zero-service approach from AWS is excellent. They just get out of your way and let you get on with things, which from the developer and project’s perspective is a massive benefit. It also happens to cost less.
Because AWS doesn’t have to be cheaper than buying your own servers. It just has to be cheaper than buying servers from you, the rabidly inefficient and slow internal IT shop. The cost structure of internal IT has so much fat in it, it would be amazing if AWS couldn’t get margins of 17% to be honest.
Incidentally, the fat in the internal IT cost structures is one reason the newer infrastructure players flooding the market at the moment are able to get sales.
Enterprise is Where the Money Is
Why rob banks? Because that’s where the money is.
Why is Amazon adding a bunch of enterprise type features like audit trails, file storage, and identity management? Because enterprises are where the money is. Depending on who you ask, the overall IT market is worth something like $3,000bn. That’s three trillion dollars. And growing. AWS is a drop in the ocean.
Ah, but Amazon is a retail company, and while it’s a large enterprise all by itself, providing enterprise IT services is not its business. AWS has a “built for the cloud” brand-new software heritage. They don’t do mainframe, for example.
Ah, but enterprise IT does. And enterprise IT, for all its faults, has developed a bunch of requirements that are there for a reason. The ability to audit things is important if you need to prove why something happened, or who did it. Such as if you get sued, which happens all the time with corporates. You have a bunch of systems and processes to handle the boring, but vital, parts of running a business that’s been around longer than 3 years. Accounting. Payroll. HR. Systems to manage all 20,000 employees leave entitlements.
AWS is learning about all these things, and is adapting rapidly, because they have to if they want to capture some of this market. AWS, for all its alleged market power, cannot dictate terms to the world’s IT market. They can influence it to change, and they are, unquestionably, but it doesn’t all change overnight.
But you know who has a long history of serving the enterprise IT market? Microsoft.
Microsoft as the Dark Horse
I am as surprised as anyone by the recent resurgence of Microsoft as doing neat – dare I say it – cool things. But let’s not lose sight of what Microsoft is really good at: enterprise software.
Active Directory is excellent. If you want single-sign-on, you have it. There are not many (and I know of zero) decent size ($100mm+) companies that don’t have Active Directory. If you know of one, I would be fascinated to learn how they are managing logins, group policy, and all the other things that basically require AD.
Exchange is everywhere. If you need to collaborate with other people with email and calendar, odds are you use exchange. Yes, you could use Lotus Notes, or Groupwise (or whatever it’s called now) but Exchange dominates.
SQLServer is pretty solid now. Windows Server is also pretty solid, if you turn off Metro. More importantly, it’s everywhere. There are reams and reams of developers out there who program boring software for corporates using the Microsoft stack. Visual Studio, by all reports, is a fabulous tool with a rich and useful ecosystem all on its own.
And then there’s all the business applications that run inside Excel. Word still dominates. Your choices for presentations are basically Keynote or Powerpoint, and guess which one corporates tend to standardise on? Visio is fine.
So while the infrastructure of specific applications might be on containers in the cloud, 90% of the business is run on Microsoft software, and has a massive ecosystem of everyone else’s software running on it.
Microsoft has its faults, but they know enterprise software very, very well. Combine that with a massive cash reserve and an existential motivation to invest in cloud, and you have a very credible competitor indeed. Throw in Microsoft’s ability to make stuff happen, the way they have with Azure, and there can be little doubt as to why AWS is scrambling to add enterprise features to AWS as fast as they can.
Microsoft nearly missed the Internet, back in the day, and pivoted so hard and fast that it surprised us all. They’ve done this before.
Ignore them at your peril.