The Swinging IT Pendulum: Thoughts on Where We Are and Where We’re Going

The technology stack is a dynamic, highly interconnected organism where changes in one part reverberate throughout the entire technology value chain. Historically, IT has ebbed and flowed between centralized vs. distributed design and proprietary, verticalized vs. open, heterogeneous architectures. In the last several years, we’ve seen a rebellion against the tech old guard – those companies which won in the client/server era with proprietary infrastructure sold through rigid perpetual licensing models (see: Microsoft, Oracle, etc.). Cloud, mobile and open source have fundamentally changed the game, disrupting the way technology is developed, adopted and deployed thereby depressing barriers to entry, innovation and adoption. Consequently, proprietary, verticalized tech stacks have, in recent years, given way to open, flexible, heterogeneous architectures – there is no disputing this phenomenon. However, during the Cloud Services break-out session of the #digHBS Summit, Steven Martin, General Manager of Microsoft Azure, confidently advocated the view that the pendulum is steadily swinging back towards verticalized solutions. I wasn’t surprised by the perspective, given Microsoft’s past and future has and does depend on a one-throat-to-choke IT model, but it did get me thinking about what’s the next evolution of IT beyond cloud.

The tech sector has always been characterized by discontinuous innovations – that is, innovations which are not built on top of existing standards or infrastructure – that give rise to entirely new markets each supported by unique value chains that standardize and then coalesce around one dominant player. Semiconductors, PCs, relational databases, local area networks (LANs) are all examples of such innovations that spawned the modern day tech giant – in the case of the innovations cited above those corresponding giants would be Intel, Microsoft, Oracle and Cisco, respectively. According to Geoffrey Moore, many of these innovations followed a predictable adoption cycle that informed product, sales and marketing and financing decisions.

Figure 1: Technology Adoption Lifecycle

Winners in this model were afforded tremendous competitive advantages as they were able to erect seemingly insurmountable barriers to entry and enforce punitively high switching costs on their customers and the entire value chain. However, around 2006/2007, the model started to change dramatically due to three prevailing forces: SaaS, mobile and the proliferation of open source software.

These three forces – in conjunction with customers’ growing wariness of closed, proprietary architectures and increased sensitivity to vendor lock-in – have fundamentally reshaped the technology stack. There are now seemingly infinite permutations of applications that can be built from choices in underlying infrastructure, operating systems, databases, application servers, programming languages/frameworks, developer tools all bound together with middleware and management tools from your vendor du jour (or open source variant). What that has created is an increasingly open, highly heterogeneous and highly complex stack or as a friend who works as a network engineer recently put to me: “sh*t soup.” This has catalyzed the formation of dozens of companies selling point solutions competing for an increasingly smaller sliver of IT dollars within the stack. Because of this very phenomenon we’re seeing the rise of what’s being coined the “full stack startup” – bundling a core technical innovation with value added applications and/or services (see: New Relic expanding into analytics). More companies are now integrating vertically to grow customer share of wallet and this is certain to have ramifications up and down the tech stack.

The question now is at what point does IT sprawl become too unwieldy to handle and when, if ever, do we see a renaissance in proprietary, verticalized solutions? In the last few decades we’ve seen shifts in the complexity of the underlying IT stack mirror evolutions in computing architecture, so if the cloud is characterized by openness and flexibility and we’re inevitably heading back towards a more closed, proprietary stack, what is the next phase of computing beyond cloud? And if the cloud was, again, about centralization of IT resources delivered through open, flexible architectures, will the next paradigm shift bring about a decentralized IT model (that we might be seeing today via blockchains, distributed compute resources and mesh networks) built upon proprietary, verticalized stacks?

These are the questions I’ll be wrestling with for the next several years. Any thoughts?