Andy Jassy, CEO of AWS, notes that, in 10 years time, most firms won’t build their own datacenters, unless it’s truly core to their businesses. Perhaps, it’s 10 years; perhaps, more; perhaps, less. Either way, the trend line is clear.

Of course, hybrid environments are likely to exist for the foreseeable future, as migrations take time, further education is required and not all legacy workloads will port quite so easily, if at all. The most prevalent hybrid cloud model, however, is apt to be a mix of on-prem infrastructure and public cloud; not private cloud and public cloud. Ultimately, private cloud is simply unsustainable, as the economics, scale, capabilities and competencies associated with building such an environment are not in rich supply at any one enterprise.

Further, if you believe that the major CSPs have more robust security capabilities, today, and that the gap between public and on-prem capabilities will only continue to grow, the justification for private cloud is even more questionable. The amount of capital, both financial and intellectual, invested in security by the largest CSPs simply dwarfs what any individual enterprise can do. While this does not obviate the need to ensure robust security architectures, policies and credentials, the notion that security concerns remain an obstacle to public cloud adaption is, today, largely unfounded.

Vendor lock-in remains a concern, and one that is not entirely unreasonable, but attempting to create an abstraction layer to homogenize all cloud access is not an easy exercise. Worse, it typically comes at a cost of reducing an otherwise rich set of capabilities to the least common denominator, while further reducing the potential economic and productivity benefits. Of course, there are tools, like Docker, Cloud Foundry and Kubernetes, that can enhance portability, but they have limits; nonetheless, it should be noted that most of the major CSPs also support these tools. In his comments, Jassy raises some good points, too, that the AWS business model is not based on long-term commitments, but rather on sustaining and winning business, every day. This is generally true for most CSPs. Moreover, APIs for various functions across different CSPs are often syntactically compatible, with AWS often serving as a de facto standard. This serves as a mechanism through which competing vendors can attempt to capture market share from one another and, in the process, provides some level of portability across CSPs.

Finally, I like to think of any cloud as having multiple tiers: core IaaS, core PaaS and enhanced/proprietary PaaS. While there are certainly some differences with things, like security models, the first two tiers cover the core set of minimally viable capabilities and are therefore largely portable across CSPs. The third tier represents a more differentiated and proprietary set of features, like NLP, AI/Cognitive, etc. There’s clearly tremendous value in these capabilities, but it’s also where one is most susceptible to vendor lock-in. Here, users must make very conscious decisions that weigh the risks of vendor lock-in with the features, pace of innovation, business strategies, competitive landscape, and time to market. If the benefits outweigh the risks, then it’s clearly worthwhile to exploit this third tier of capabilities; if not, then it isn’t.

Either way, while public cloud adaption was initially limited to the most innovative, embryonic or risk-free use cases, that is certainly no longer the case. Businesses of all sizes, across all verticals, are increasing their uptake, including many of the largest, most highly regulated, entities in the world. For smaller players and upstarts, of course, public cloud has dramatically reduced capital and time-intensive barriers to entry, resulting in increased competition for legacy incumbents. In industries, like financial services and retail, where margin compression continues to accelerate for a variety of reasons, including, though not limited to, the proliferation of new market entrants, public cloud not only represents an economic arbitrage opportunity, but is increasingly a parity play.