Fun with market sizing

I’ve spent a lot of time over the years working with cloud market-sizings, and occasioanlly on them. They’re always a bit whackadoodle and can be difficult to pull apart. But, so long as they’re consistent year of year, they do give a good intedication of momentum and a comparision to other markets. This is what you should be using emerging technology marketsizing for: just indications of which way the wind is blowing and how strong that wind is relative to other breezes.

All too often, strategy and M&A people (and other “MBA” types who’re doing valuations and finances, plus all the hangers on in the chattering classes) get obsssed with market-sizing as if they’re “real” and start to do things like include them as fundamental parts of their business plan, e.g., “we’re going to capture 1% of the cloud market this year!” The implication there is that if they don’t, they’ve failed…but when you realize how corny the market-sizing are, you realize that basing your yearly plan on an Excel macro is a poor use of time.

With that disclaiming context established, I love finding market sizing numbers. They’re fun once you get enough of them and can start figuring out the relative size of markets. Which is to say: how much money is being spent each year across the world on various types of technology.

PaaS

Platform-as-a-Service is one of the more tricky markets to size and has spun all sorts of directions over the years. It’s always the smallest of the three aaS’s, but the highest growth. Even worse, most PaaS market-sizing you see is for only public PaaS. I’ve heard that Gartner has consternated much about this over the years: if you insert a simple “r” into it, they have an on-premises category called CrEAP that sizes what, I think, is “private PaaS,” and I believe they’re fixing it up further.

However, the problem with sizing the PaaS market is asking what you’re sizing. There’s “PaaS from SaaS” offerings like Force.com and then so called “first generation” PaaSes like Heroku (also at Salesforce) and EngineYeard. Then there’s hosted development tools (all the CI services otu there) that, for some reason, show up in PaaS market-sizing. But now there’s all the private PaaS offerings that happen to also run as public PaaS (it turns out the enterprise market really likes private cloud as well as public). And then there’s trouble-makers like us at Pivotal who bristle at the notion of being called (just) a PaaS. Layer Docker, Mesos, kubernetes and all those folks in there…and you’re head should be spinning.

The composition of this market has changed dramatically over the years. My theory is that it’s not well “shaken out” (defined) yet and it’ll change more. So, when I get asked for PaaS market-sizing, I sigh a bit inside.

Getting to a helpful PaaS market-sizing

I think the best process is to thinking through a what PaaS is used for and then try to figure out how much money is spent on solving that problem, not the exact technologies used to solve it. To me, that gets to some indication (whether it’s a ceiling, floor, or mid-point, I don’t know) of how much money there is up to grabs if you’re selling PaaS.

To that end, I always like this chart from a recent Goldman PDF:

As you can infer from my over contextualizing above, I think most PaaS market-sizing is bunk, but this chart a good way of thinking through getting to answer. It compares traditional packaged software and on-premises hardware spend to IaaS and PaaS to show how it starts to erode into “non-cloud” IT. I’m not sure if IaaS and PaaS and public cloud only (it probaly is, which is problematic).

When using this chart, the voice over is something like “tracking this market is difficult at this moment, as any analyst will tell you. I like to use something like this chart as a guide for thinking about it. With PaaS, what you’re interested in is how much traditional IT teams writing cloud-native applications are taking over, and this is one swag at it. Notice that the growth rates are drastically different too: there’s little to no growth in traditional.

The other thing (since most analysts don’t track private PaaS) this suggests is that all the traditional middleware money switches over to all PaaS (public and private) at some point…at least all the “new” money, the growth. Or, at least, that it’s a rough heuristic. It could be less (prices go down with cloud, right? ;>) or it could be more (software eating the world X IT - SaaS = what? == more customer software development at companies leading to more spend on the application development category).

So, if you add up the traditional markets of Appdev and middleare, you get something like a $35-40bn market in 2018 or so, depending how exuberant or dower you want to be, for pubic and private PaaS. Again, that wet-finger-in-the-winding is “bunk,” but it tells you type of money you should think about. Assume that over the next 10 years most “appdev and middleware” spend converts to “PaaS” and then you’ve got something that looks less shitty than the PaaS market-sizing analysts do now-a-days…and more real, I think.