One hundred years ago, there was no electrical infrastructure. Edison and Tesla had not begun their battle over whether we should use AC or DC to distribute electricity, and lamps were still being fueled by the dwindling resources of the whaling industry. Railroads had just agreed to a standard-gauge track so that trains could cross from one line to another. There was no interstate highway system. For that matter, there were no mass-produced cars.

In the short time since, we have gone through successive stages of innovation followed by standardization and commoditization. In the utility space, each fueled an increasing availability of power to our homes and offices. We used to buy refrigerators and then go out and select the motor to run in it. Now we are ignorant of what’s in the box and where it comes from. Such is the way of industry.

IT is no different, with successive waves of innovation bringing us a standardized and commoditized offering. Computers used to be stand-alone and unique. My first job entailed getting computers of different makes to talk to each other. Largely thanks to the Microsoft monopoly, the hardware commoditized, which allowed the networks to standardize. Thanks to DARPA, we have the Internet, which gave the machines a way to use those networks to interconnect, pushing us past the tipping point to where we are approaching ubiquitous access to computing power. The tech bubble left us with gobs of dark fiber and comms capacity just waiting to be used (and driving down the cost of interconnection).

But IT is different from electrical, cable, and telephone utilities. IT can be anywhere, and the others are point-to-point, i.e., you need to know where the central office is. They are essentially N:1 connectivity, like a arms radiating from a central hub. The Internet means computing power, and by association the content you access can be anywhere, i.e., it is N:N connectivity, where the N in this case is in the billions. Hence the concept of the “Cloud.”

A couple of years ago, I stepped in to run a server-hosting business in Texas. Server hosting means that instead of building a data center and buying your own boxes to stick in it, you rent space on boxes in other people’s data centers. You pay for this by the drink, i.e., upfront CapEx with low NPV is replaced with variable OpEx with a high NPV, no associated staff costs, and no long-term commitment.

The clients of this firm were largely gamers, playing first-person “shoot-em up” via the Internet, and they were all over the world. You may think that gamers are a bunch of teenage slackers lacking social skills — and indeed they may be — but believe me, they have high service expectations. Online gamers expect that there is no noticeable delay between when they pull the trigger and the gun fires. Moreover, the ones in Kuala Lumpur expect the same response time as the ones in LA and NY. The firm in Texas gave them this across more than 10,000 servers on a 24×7 basis, and I bet not one of those gamers knew or cared where the servers running their games lived. That’s the Cloud.

Since that time, the technology has moved on to encompass not just the hardware, but more and more of the content we run on it. What works best in the Cloud is content and services that are essentially generic, so we are seeing such things as:

• Exchange-based e-mail

• Anti-virus

• Anti-spam

• Storage capacity with backup

• VOIP

You buy these services on a unit-consumed basis, either per-seat, in the case of e-mail accounts, or per gigabyte, in the case of storage. Little to no CapEx up front. High NPV as what you buy is immediately and fully useful, no staff costs, no long-term commitments (although you will get better pricing if you do make longer-term commitments).

I am now advising my clients to look seriously at these services before they ever consider building a data center, buying a server, or paying Microsoft another server licensing fee. The economics have changed that dramatically. Moreover, the pricing and packaging of these services have become meaningful to the corporate consumer. We have moved from needing to build an IT department, kit out space, buy servers, and install Windows in order to get our e-mail to a simple $20 per-seat-per-month charge. Anyone who buys cable TV for their home can understand that.

And we are in the early days still. Edison and Tesla are still wrangling over dominance, we are still laying the intercontinental tracks, we still need to buy tires from the company that made the car we drive, but we have reached the tipping point and things are moving fast.

We are beginning to see viable offerings of business applications over the Web, starting with SalesForce.com for CRM. Notably, Oracle has added a Web-delivered ERP solution and several GL offerings are either out or coming. You are beginning to hear of them under the rubric of Software as a Service (SaaS). You will buy them by the drink and consume them through any computer with a browser. Think of your browser as a TV set on steroids – it doesn’t do much by itself, but through its standard interface you can access an infinite variety of content.

The greatest challenge many companies will face is not in accessing and using these services, but in getting around the sunk-cost nature of their established infrastructure in order to migrate to this new paradigm. Companies that step up to the plate will be positioned with a much lower cost structure that scales more quickly when the economy opens up. Certainly no firm should be opening new operations without looking at the Cloud.

Is it all that easy? Of course not. You need to worry about where your data lives and the security and process around it, but this is a more appropriate focus than worrying about changing the oil in the engines that house that data. You need to ensure that your providers’ data centers are SAS70 and that you connect to them through secure linkages. You need to make sure there are appropriate backup and disaster recovery procedures in place and that they are followed. There may be HIPPA and ERISA issues that need to be accounted for, and so on. In other words, cost goes down and service goes up, but the management doesn’t go away.

Yes, there will always be a need to run your own data centers and computing. Many companies run their own power plants. Data may be too sensitive to not house within your walls. You may have such a big portfolio of proprietary applications that you end up with the staff and facilities anyway, making delivery of office services no more than a marginal cost. But the Cloud has the potential to create a discontinuity in how companies consume and pay for services, and every company should be thinking about it today. ###