Cloud is to the Application what Virtualization was to the Server

Cloud is more a business model shift than a technical shift.  By business, I’m not speaking of financials and the respective trickle down savings and efficiencies but instead a seismic shift in how IT services are consumed, paid for and sold. Part of the results of this impact can be seen in Wall Street and not through only the usual suspects of Server companies but, more interestingly, through the revenue patterns of any company selling software.

To illustrate this I would like to reflect back to the proliferation of Server Virtualization from the 2000s.  The technical impact was obvious and necessary; due to Moore’s Law we found an oversupply of processing power and no application could be written to fully use that power.  Server Virtualization was an excellent tool to drive up utilization of capital assets such as Servers and with it came some incredible benefits.  Because of the level of abstraction from the hardware and the ability to affect change at the software level without the physical movement of hardware this gave superhero abilities to Server administrators. Along with higher utilization rates, making the bean counters happy, additional IT services became easier, faster and cheaper.  Disaster Recovery was suddenly available to all applications not just the ones running on multi-million dollar machine ecosystems.  The speed in which servers could be deployed went from weeks down to hours.  The most important factor is that Server virtualization did not significantly change the way infrastructure looked or was administered.  Similar to Virtual Tape Libraries, even though there were no actual tapes being written to from the Mediaserver the abstraction layer still provided the same look and feel of tape.

This is where Virtualization and Cloud diverge.  Virtualization, while amazingly transformative and something which Cloud is built upon, did not fundamentally change the way IT services are consumed, paid for or sold. 

Similar to how there was an underutilization of processor capacity in Servers, we are finding similar levels of underutilization in Applications.  Microsoft Office is a prime example of this.  Don’t get me wrong, it’s an amazing product but we should ask ourselves, ‘how much of the product we use vs. what we pay for?’. Word is equipped with advanced publishing tools suited for best selling authors and Excel is able to compute complicated financial and math problems something which a PhD in Geophysics could only come up with.  This divide of what we get and pay for and what we use is called the Consumption Gap.

Looking at Enterprise Applications, we are seeing a similar Consumption Gap.  There has always been a decent level of customization around Enterprise Applications covering Logistics, CRM or ERP however that customization came at a cost to the customer.  To be fair there are software-licensing tiers which cater to different business and budget needs however the customer is always burdened with the risk of project success not the vendor.

When deploying a new Application, in the traditional model, the customer will need to procure an infrastructure to host the application. Virtualization has done a lot to cushion the upfront and operational costs associated with this but the software license for the given Application and its auxiliary dependent services is an upfront cost and depreciated with a service and support contract providing additional cost over time.

What this creates is a model where the customer, not the vendor, is on the hook to realize the ROI for the given Application project. Whether the application is a success or not, the customer is holding the bag.  Let’s not forget about the consumption gap between the features being provided versus used.  This adds additional ROI strain.

Lets take an example of Nuance Technologies, a company taking voice-to-text solutions to the next level.  One of their solutions is positioned towards doctors who are freed from the burden of typing up patient notes by instead speaking the notes into a small hand held device which then translates the speech to text in the cloud and places it into the respective patient record system.  This is done through advanced speech recognition and self-learning artificial intelligence to better decipher the doctors unique word pronunciation. Nuance recently made the shift from a license software model to a subscription, pay as you go model. Initially, as you can imagine, this translated into a significant hit to their quarterly revenues because no longer were they receiving a 3 year upfront payment for a software licenses but instead a smaller payment occurring on a monthly basis.  I personally believe that this will benefit Nuance significantly in the long run however the shift in business models gets Wall Street on edge. This change is being negatively reflected in the short term stock price.

What Nuance is doing is where the industry will eventually go.  This is irrespective of whether the software is hosted by a Cloud Provider or is deployed internally on premises.  Public Cloud or hosted Cloud is not the default in this model, it is one of many options.  One thing which is the same, is the underlying business model will have shifted and with that the risk to the customer.  Software as a Service, is enabling customers and vendors to close the consumption gap while shifting the ROI risk away from the customer and towards a shared burden with the vendor. 

Business models are difficult to change and with that will come turbulence, there will be some significant winners and for the companies who don’t evolve there will be some losers. This does not mean that license models are dead but it will prove out that some significant form of innovation would need to take place if a consumption business model is not adopted.


Great example with Nuance - a conclusion could be that the new business model is easier to adapt for non-listed companies ?!


Tim, good analogy with utilisation on the server level, I certainly agree. I guess the more challenging and interesting part will be the revenue models that you point out by the example of Nuance. Existing Cloud Service providers usually do run the recurrent subscription models already, so there is no challenge for them to report on it properly. The bigger challenge will be on the vendor side. If the future is all about consuming applications and infrastructure from the cloud, those systems need to be build and might need a different licensing model also on the vendor side. Once this is happening the whole revenue and compensation model on the vendor side needs to be adjusted, similar to insurance sales p.e.

Also the contractual part will need to be reworked and new financial reporting processes will need to be set up.