Who gets killed by cloud computing?

"It's only a flesh wound!!"

It's always more fun to watch someone get cut off at the knees.

As I mentioned at the end of my post on Tuesday, as well as winners there are big losers from Cloud. This is because a shift in IT architectures shifts where IT budgets are spent. This change in spending can rapidly change the fortunes of IT vendors.

In fact this is one of the big themes of my blog: How giant tech empires can be disrupted almost overnight, while other great powers rise in the blink of an eye.

We can vividly see it in the world of cloud computing. To understand why we need to look out how Cloud works in practice.

It's not rocket science

Cloud isn't rocket science. It really isn't. As I intimated before, its basically old-school business outsourcing in a frock.

The chart below shows the typical IT set-up in a big business. In particular look at the purple column on the right. This shows how an IT system is built up - layers of different functional systems (e.g. server, database, operating system) built on top of each other. Each layer depends requires all the blocks beneath it to work, and each layer is generally supplied by a different vendor.

In a traditional in-house IT setup, the second column, all of this is run by the end-user (the company deploying the IT). They buy all the components from the different vendors, stitch them together (with a little help from their mates at Accenture) and run it themselves.

What happens in different variants of a cloud offering is that more and more of these parts are run by a third-party outsourcer, the "Cloud Computing vendor":

There are three main types of Cloud Computing:

• Infrastructure as a Service (IaaS): I always describe this as like a homeowner buying a shell of a house - four walls, some windows and a ceiling but nothing else. In real terms this means that the cloud vendor runs a data-centre full of servers with basic functionality, and lets the users install whatever databases and software they want on them. The most famous exponents of this is Amazon, which leases out space in its Elastic Computer Cloud (EC2) for a pretty low sum. 
• Platform as a Service (PaaS): If IaaS is like buying a shell of a house, Platform as a Service is like buying a house with the electricity and the plumbing already laid in.  Instead of just running a dumb server, the cloud vendor also slaps on a database and an application framework. You have APIs which can take care of mundane functions such as connecting to a printer, or authenticating a user when they log in. But you are free to build pretty much whatever app you like on top of this. Salesforce.com runs a platform called Force.com which is the increasingly common in enterprise computing. Zynga games such as Farmville running on the Facebook platform are a consumer equivalent.
• Software as a Services (SaaS): This is like renting out a house already furnished and connected up to the utilities. The vendor builds the whole end-user application and you just access it, typically via a web browser. GMail or Hotmail are the most commonly used SaaS applications. The website you're reading this on now - Blogspot - supplies all the tools needed to create a blog as a SaaS offering.

Note how this relates to the chart above - at every stage in the evolution from Infrastructure to Platform to Software as a Service, more and more of the traditional tech stack is being outsourced to the cloud computing vendor.

Also note that a lot of this isn't actually that new. Infrastructure as a Service is simply traditional infrastructure outsourcing in a frock. Software as a Service is hardly new either - Hotmail.com is sixteen years old. What is new is Platform as a Service - that is a genuinely innovative way of doing things that has the potential to change how software is written.

Cut off at the knees

So far so good. But note the darker side of this for the technology vendors.

As each layer of the IT stack is taken over by a cloud vendor, the companies which supply that component are screwed. 

Think about the market for databases. Oracle has a lovely businesses selling relational database software to millions of businesses, all of whom have exactly zero negotiating power over Oracle. However once you move to a Platform-as-a-Service or Software-as-a-Service model the end users no longer buy the databases. There is only one database buyer in the market - the guys who run the PaaS or the SaaS. So at best Oracle's negotiating position is dramatically weakened (they have to deal with a much larger customer). And at worst the cloud provider might decide to junk the industry-standard Oracle database entirely and build their own data management layer. In fact this is precisely what SAP is doing with its Business ByDesign cloud ERP software - it runs on an in-house database evolved from SAP's own MaxDB product.

We are seeing a similar effect going on in servers. As I wrote before, Google have long built their own servers and other cloud companies like Facebook are following suit. Instead of going to traditional server vendors such as Dell or HP, Facebook is pushing the Open Compute initiative to build generic low-cost servers. Server incumbents lose out, components suppliers which might sell motherboards or memory into Open Compute servers win out.

Think about IT Services companies like Accenture - traditionally much of their profits have been derived from bread-and-butter systems integration work, making legacy IT systems all work together. But if we move to a SaaS world there is no longer any legacy IT to stitch together; it is all run on Google's already-integrated Apps platform.

And what about Microsoft? How much money will they make selling Office and Windows if I just pop up Google Apps in a Linux browser and start tapping happily away.

Its interesting to see Sony buying cloud gaming vendor Gaikai - because if broadband networks catch up, how much demand will there be for a power-hungry home games console if you could just stream Halo XVIII frame-by-frame down your fiber connection?

Everywhere you look there is potential for dis-intermediation as the world moves towards Cloud Computing. Of course that is not to say the empires aren't striking back. They still have vast financial resources to compete against cloud disruptors or to simply buy them out before they become a threat (a favoured tactic of Larry Ellison at Oracle). And certainly in enterprise IT world (but less so in consumer tech) customers are incredibly conservative and slow to change. Hey remember both business are still on 2001-vintage Windows XP!

But if you want to see disruptive technology and business disintermediation going on before your eyes, just look to the Cloud.