Utility-Scale SaaS: a Shift in the Enterprise Software Market

SaaS as a Utility


Enterprise software has been experiencing a transition towards “cloud” models of various flavors over the last fifteen years. As technology vendors, buyers, and advisers this transition has involved an evolving sense of how and why to embrace cloud. In effect, the industry has been going through a protracted and decentralized conversation about the definition, benefits, and drawbacks of “cloudness” in order to answer the simple question of “when is cloud a good idea”. This article tries to provide a very brief recap of how that dialogue progressed up to today and then to speculate on where it will go in the future.

Defining Cloud

Cloud, at its simplest, indicates software that is not controlled within the hardware or network of its user company. It is “outside the four walls” in some sense. But there are ways a software can be more or less “cloud”. Aside from its location of installation, a software’s “cloudness” essentially relates to two factors of its design: the extent to which it is delivered as a service instead of an asset and the extent to which it enjoys economy of share based on software usage. These two factors matured separately and cultivated separate value proposals. Software as a service has key benefits in terms of IT economics in many (but not necessarily all) situations. It converts capital expenditures (CAPEX) to operating expenditures (OPEX), an act of financial engineering of value to many public companies. This financial (and process) engineering has much in common with the generation-long trend of outsourcing business functions to create asset-light, core-competent companies. Software as a service also enables alignment of cost-to-usage, and to provide faster return on investment because less up front investments are made. And by lowering the absolute magnitude of one-time investment, it enables smaller scale deployments either in service of smaller business needs or a pilot prior to full scale rollout.

Meanwhile, the second dimension of “cloudness” is the degree to which it enjoys economies of share across its usage base. An economy of share is simply a way of saying more value is created by having more users. Elsewhere this might be called a network effect, or a positive feedback loop. It is the engine of growth behind consumer software like Facebook and Linked-In, but not very well known within Enterprise software. Economies of share can come from any layer of the process and technology stack for a given software. At the lowest levels there are pure overhead sharing: a single data center with shared staff. Then perhaps is shared hardware of some sort: the same server rack but with different slots for different companies. Then it’s a shared hardware server but running unique instances of software. Then it’s the same software instances, with separated databases. Then it is shared databases such that even the information is a collective investment. Then it might be named users that are shared, and so forth. The more economy of share taking place, the more a software could be said to be “cloud” because it’s moving away from the ownership of a single company and towards a community-management model.

Clearly there is some degree of reinforcement here: it would be hard to find software as an asset that leverages high economies of share. But it’s not impossible, and I stand behind separating how a software is paid for and to what degree its driving components are shared as two drivers of “cloudness”.

When is Cloud a Good Idea: Past and Present

I really didn’t want to write an article about how to define cloud software, but it seemed like a necessary first step to postulate a definition. Now let’s look at the question that the enterprise software community has been promulgating for fifteen years or more: when is cloud a good idea? When it first appeared as a technically viable option, most of the conversation revolved around reliability and security. To give an example, I heard many firms such as IBM and Accenture advising in the late 2000s that only non-essential (due to unknown reliability) and low-sensitivity (due to unknown security) business use cases should be migrated to cloud software. These sensitivities to reliability and security were both accurate and short-sighted. The technology and commercial models for cloud are no more inherently unstable or risky than procuring and operating software inside a company. It should seem obvious that with time these roadblocks to fuller adoption of cloud would go away. Then what? It seems we’d be back at the same question: when is cloud a good idea?

Fast forward to the current day, end of 2014. For several years the advisers who had once constructed their guidance around reliability and security now are most sensitive to IT economics. In other words: some use cases will have better return on invested capital via cloud and others through on premise software-as-an-asset. My commentary on this view is that it represents a kind of “level playing field” view of the software space. Cloud is neither a magical elixir nor is it necessarily less robust or safe than in-house software. It’s simply a different software acquisition model, and that difference is easy to evaluate in the straightforward math of ROI. As an example, a company’s ERP software was once considered too critical and too sensitive to be in the cloud. Now, a company may have its feature-set covered by a cloud offering like Workday as well as an on-premise solution such as Microsoft Dynamics. The final decision isn’t then made based on reliability and security, but on ROI.

When is Cloud a Good Idea: the Future

After setting the background with a definition of cloud and by how it was and is currently selected as the right technology approach, let me now move to speculation about the future. As I mentioned above, the prevailing view is to treat cloud and non-cloud software as existing on a level playing field. I advocated this position until recently, even including it in my book on supply chain visibility. But I think that view is inaccurate. There are some enterprise software solutions which are simply better in the cloud. The distinction, I propose, is that such software is cloud mainly because of a strong economy of share instead of primarily because it benefits financially when sold as SaaS. Selling a software as SaaS or as an asset is a kind of financial engineering (supported by operational differences of course). But, it juggles operating cost and capital investment in a way that can be good but has natural limits: it’s about slicing up a pie differently and eating it over time. Economies of share are different altogether: they alter the size of the pie.

A strong economy of share means that the more usage, the better the software is for everyone. The stronger this effect, the larger the value that can be tapped when such software is deployed via the cloud compared to non-cloud. When very strong, such software will produce winner-take-all markets. In the context of enterprise software as opposed to consumer software, successful cloud applications will achieve the scale of utility-grade infrastructure. To be specific, I’m making three independent speculations about cloud software for enterprises:

  1. It’s true that cloud software has overcome reliability and security issues. It’s not true that this means it is on a level playing field with non-cloud software: their differences go beyond the cost of ownership economics.
  2. Software takes two paths towards “cloudness”: via being offered as SaaS or via achieving economies of share from its usage base. The first is a dead-end, even if valuable. The second is revolutionary: it’s new to enterprise software (even if well known in consumer software), and it produces software that is massively superior when delivered in the cloud vs. in-house.
  3. As enterprise software emerges or is engineered to maximize its economy of share effect, so will these softwares produce winner-take-all markets closely resembling utilities underlying industrial economies.

If these speculations prove true, the impacts will be far reaching. Although I haven’t solidified my thoughts about where the market would go, I could raise some questions. For example, will this result in increased cross-pollination between consumer-software and enterprise-software talent? Those worlds are nearly completely divorced today. Will the giants of consumer software (Facebook, Yahoo, Google, Linked-In, Ebay, etc.) cross in to enterprise software themselves once they see the ability to deploy what they already know about viral engines of growth and cloud software engineering? Industrial utilities such as water, heat, and electricity are often awarded time-limited monopolies… could this occur within enterprise software markets? How will market dynamics of product engineering, launch, sales, and acquisition be altered by the presence of winner-take-all markets where network effects predominate other factors? If it’s nearly impossible to launch a new Facebook, does this indicate someday it may be nearly impossible to launch a supply chain finance portal (because 90% of possible participating companies are already active on the incumbent)?


Software is called “cloud” when it is either: (1) sold as a service delivered over time instead of a one-time asset purchase, or (2) it enjoys economies of share based on a broader usage base. In the last ten years the enterprise software community (its vendors, buyers, and advisers) have spent considerable energy in a decentralized discussion about “when is cloud a good idea?”. At first, the consensus expert opinion was that we should create, sell, and buy cloud software mostly when the reliability and security needs were low. Hence it was deemed unstable and risky. Now cloud seems to have outgrown these perceived drawbacks. The predominate view today is that there exists a “level playing field” between cloud an in-house software deployment models. This view states that the two models are about the same and that some usage cases will have a lower cost of ownership on one format or the other. It’s mostly a decision made by ROI for the buyer (and hence a decision of pricing for the vendor). What this article suggests is that it’s time to abandon the focus on a “level playing field”. Although ROI & pricing are critical, the guiding principle behind when cloud software is a good idea should acknowledge its special ability to tap economies of share. Strong economies of share, whether inherent in the business need or engineered by the vendor, lead naturally to winner-take-all market dynamics. In the context of enterprise software, I suspect it also leads to utility-scale software: modern equivalents to industrial utilities for industrial economies. Where that then leads is anyone’s guess…

Leave a Reply