• Will cloud computing reduce IT budgets?

    All business activities commoditize – that is, they evolve through a common pathway from their innovation, to custom built examples, productization and eventually commodity provision. It is in this latter stage that utility services can appear. IT is no exception to this evolutionary pathway and cloud computing is simply a popular term to describe a wide range of IT activities that are currently evolving from products to utility services.

    This evolution creates a wide set of beneficial changes which are directly associated with efficiency, from economies of scale through volume operations to a tighter linking between expenditure and consumption. A superficial examination of this can easily lead to an erroneous conclusion that these efficiencies will result in reduced IT budgets.

    The problem with such a conclusion is that it ignores demand side effects and competition:

    • On the demand side, most companies have a long tail of unmet demands on IT and efficiency gains can simply be consumed by doing more. Any examination of price elasticity in IT over the last 30 years will show that this is the general case, as computing has become more efficient then increased consumption has negated any potential savings. A consequence of Moore’s law is that $1,000 will buy a million times more computing capacity today than it did in 1980, but during the same time IT budgets haven’t drastically reduced due to this increased efficiency – we've simply done more.
    • On the competition side, a consequence of commoditization to utility services is always increased rates of innovation for new activities built upon these standardized components. Historically, as our competitors have created new activities in IT (from internet real estate to social media), we’ve generally had to respond. IT is an arms race and we are all caught in a game of running just to stand still relative to one another.

    It is these two factors of demand side and competition effects which will tend to outstrip any supply side efficiency. Whilst the breakdown of an IT budget is likely to change with a smaller fraction allocated to maintaining existing environments and a greater portion spent on new activities, our overall IT budget is unlikely to reduce. Unless, of course, we can somehow prevent competition, innovation and our desire to consume.

    Examples of this paradox of efficiency causing increased consumption can be seen in many industries; electricity consumption exploded as it became a commodity provided through utility services. However, to give a fair historical perspective, this effect was first noted by William Stanley Jevons in 1865 when he showed that technological progress that increases the efficiency with which a resource is used, tends to increase the rate of consumption of that resource. In his day, the steam engine had been made more efficient and a general observation was made that this should lead to less consumption of coal. Stanley showed that the opposite happens.

    There is no reason to doubt that Jevons’ effect won’t hold true with cloud computing. You shouldn’t therefore expect to reduce IT budgets through cloud computing – you should simply expect to do vastly more.

    Recent press/video articlesFollow this blog:

    Wall Street Journal Europe (February 2011): Pinning Down Cloud Computing.

    Watch one of Simon's presentions on YouTube from July 2010, entitled Situation Normal, Everything Must Change

    IT Business Edge (July 2010): Maintaining Strategic Control over Cloud Computing

  • Post Comments

    Discuss this Blog Post

    1. Simon Wardley says:

      Hi Kevin,

      Yes but I'd argue it's a little more subtle in impact.

      Services can appear in a product world e.g.rental but they are often tailored to the user, highly customisable and based upon some element of time sharing a high cost product. In IT they've also tended towards solving capacity and resilience issues through physical architecture e.g. bigger machines, N+1.

      Utility services are all about volume operations, standardised good enough components and commodity provision. In IT they tend towards solving capacity and resilience issues through software architecture e.g. distributed systems and design for failure.

      The two are based upon different economic models. This is the key issue with cloud computing, commodity-like activities (those which are ubiquitous and well defined) are being treated as commodity-like activities provided through utility services. Cloud has this model at its heart.

      Certainly an awful lot can be learned from supply chain management techniques and applied to cloud but I'd argue that the future of this industry is wrapped up in the concept of commodity provision and not just simply a better mechanism of balancing existing demand through a central service.


      Simon Wardley

    2. Simon Wardley says:

      Hi Alex,

      Excellent comment which opens all sorts of different areas of discussion - it's several posts worth alone. However, for now, I'll just expand on one point you've touched on - the role of government and regulators.

      Any evolutionary shift, whether from custom built examples to products or from products to utility services, brings a common set of risks to consumers. These are :-

      Disruption risks to an existing model, including loss of political capital, changing relationships with suppliers and loss of previous investment.

      Transition risks from moving one model to another including trust, transparency, governance and confusion.

      Outsourcing risks including pricing competition, vendor lock-in, loss of strategic control etc.

      Whilst there are many standard supply chain techniques (e.g. use of markets, use of hybrids) for mitigating several of these risks, the governance issue is impacted by government legislation and the operating environment of the company.

      Now along with the often quoted benefits of cloud, the shift from products to utility services will result in accelerated innovation of new activities built upon these services, disruption of some existing product vendors who are unable to change to a utility model and reduction of barriers to entry in orthogonal industries (i.e. ones where access to large volumes of compute resources has been a barrier beforehand).

      This has quite a large societal impact, and so it's no surprise to find the Chinese government investing $7.5 Bn in cloud computing to build what is dubbed as "cloud computing valley". This isn't being done because of a desire to make cost savings, it's naked competition and a dash to become the future focus of this change.

      But then, this shift to utility computing is all about competition whether we're talking companies or nations. The problem with companies not being able to exploit this change due to local regulation is the impact is not limited to those companies, it has a far wider societal implications.

      However, in equal measures, regulation can be used to encourage a local industry. Was China's censorship of Facebook simply to control social media or was it used to enable local services (ReRen, TenCents, Baidu etc) to develop in order to avoid a future where the social network was in effect controlled by a Foreign power? There are all sorts of strategic issues here at the national level.

      But that's the point. Regulators need to be looking at cloud computing and working out how to game the environment and encouraging development of the industry in their best strategic interests. Those who aren't, are allowing other nations to take an advantage.


      Simon Wardley

    3. Kevin Neifert says:

      Sometimes I wonder if we IT folks think we're inventing something new when retailers have been using warehouses and centralized distribution centers to manage the ups and downs in demand for years. Isn't cloud computing really just a distribution center model for IT services?

    4. Alex Winch says:

      I couldnt agree more Simon, however in another point, my experience of cloud so far is that the rules and regulations arent allowing companies to make best use of the cloud anyway. Cloud is still a great opportunity to manage overflow and impromptu demand but whilst legal challenges of US access to data continue to drive agendas, full exploitation is limited which means cloud becomes another cost (adding value sure) but not reducing a cost necessarily from somewhere else as either duplication or limited engagement prevents that exploitation. It still feels like people are waiting for the cloud equivalent of service pack one to make the best use. Additionally you could argue that companies have already invested in the infrastructure to run their business now, so cloud is something that is gathering momentum around the periphery but hasnt been fully adopted whilst assets are still being sweated.