Moving to the Cloud is Creative Destruction

A recent article in "the economist" magazine questioned the value of the public cloud altogether. This took place in the Schumpeter section nonetheless. Joseph Schumpeter the economist most identified with capitalism’s creative destruction was transformed into what Peter F. Drucker refers to as a bean counter when it comes to the cloud. This kind of article and thinking will reinforce Information Technology’s old guard’s skepticism of all things new and especially the cloud.

At the heart of the article is relating a blog post from venture capital firm Andreessen Horowitz using the example of how Dropbox managed to save billions and increase stock market valuation by repatriating the public cloud resources to its own datacenters. Indeed, as the theory and studies show there are cases where private on-premises deployments can be more cost-effective than the public cloud (as can be viewed in this lecture on cloud computing from the University of Illinois and Coursera). 

The true benefit of the public cloud, as well as more elaborate models such as multi-cloud deployments, comes from the increases in agility and the fast responses to market changes. This is especially true when the enterprise implements a DevOps mindset, is cloud-native, and unshackled from legacy code.

Companies where core competencies should not include fine-grained infrastructure management, and where it does not bring them a strategic advantage should not be in the private data center business (outsource partners or not), even if it results in perceived cost savings.

Dropbox's core offering is to provide cloud storage services to end customers. They are a cloud provider by another name and as such would be at a competitive disadvantage vis a vis other providers of similar services who control their own cloud infrastructure. For instance, Google with its Drive offering can thus provide cheaper storage services to end customers. On the other hand, a public-cloud-using-Dropbox would be beholden to its infrastructure provider.

Most enterprises are not repackaging some cloud service and selling it to customers. They are offering SaaS or non-software services and thus would need to add to their core competencies data center management or deal with an outsource data center manager if they are to use a private cloud. All of it would reduce agility.

So, from the strategic perspective of focusing on one’s core competency, it would be best for those enterprises to move away from infrastructure management and outsource it to the public cloud(s) which have made infrastructure management an API away. This is in fact the same logic as to why non-software firms are best served by hiring consultants as opposed to being in the software development business "proper" themselves. Even if the latter could be cheaper.

The question is thus asked, where does this leave the much-hyped hybrid cloud in the enterprise. I would argue that it is a transitional phase to the public or multi/public clouds. Most enterprises are shackled by legacy IT systems and use private data providers. Various strategies of different maturities exist to bring them to the cloud. However, like most things in business, the bean counters need to be convinced to spend the millions today versus hemorrhaging a lot more over the years to patch systems that are written in arcane programming languages and in a monolithic and sometimes hard-to deploy fashion.

The ultimate expression of Schumpeterian Creative Destruction lies in staying agile, gaining market share, building software, using throw-away public cloud components such as functions, and thus meeting your customer's needs in real-time. Bean counters be damned.