My take on state of the Public Cloud market – what does the market look like currently, what customer segments are there, what’s driving the adoption behaviors and how would this market evolve in say next 18-24 months etc., and as I ruminated(not thought, ruminating is much deeper… actually just sounds better than plain thinking – everyone thinks, few ruminate … some words are just fascinating)… so as I ruminated, few things were clear:
I am not an expert so why bother?
I don’t have any quantitative data so what would I base my ‘assertions’ on?
And more importantly, who cares about my take?
And then I came across this IDC study… and BOOM, it all came back to me. Actually I had clearly stated at beginning of my blog writing journey that my goal is to break my keyboard by banging away thoughts as they germinate (another fascinating word) in my head… and I am proclaimed un-original thinker. Ergo, (another fascinating and fancy synonym for much simpler word) I shall keep ruminating and writing. Now, one would think how the referred IDC study could have such unintended impact. So here it – this was borne out by the profound sense of loss and pain of having paid $5 for that study, which essentially has all the 3 issues that I chanced upon my rumination above. Ok, not the 1st one, as IDC is chock full of experts… but you get the drift.
I meet with lot of customers day in & out, lot of experts as well and a smattering of smart people who are well read on wide range of topics… and since I am tad more passionate about Cloud & Mobility now a days than I have been for any other technology lately (last was when I was coding 3G physical layer algos for Nokia way back in 2001-2002), I pay attention. One word that I hear more often than others in Cloud conversations is ‘scale’ and this is the one I will share my thoughts on in this post.
What exactly does ‘scale’ mean in context of cloud? Is it same when customers/end-users say it vs. when it’s used by vendors or Cloud Service Providers (CSP)? And if it is coming up so often, then it must be important, is it so and if it is then why?
Most customers I meet seem to use scale and elasticity interchangeably – as in: I should be able to scale in and out and when I need to scale out/burst, then I need guaranteed capacity. And that’s what elasticity means as well… your increased need for compute resources will be met. This is essentially a technological/architectural point of view (pov) that’s driving the use of word ‘scale’ – apps should be coded to scale in/out vs. scale up/down (prevalent app writing/running paradigm on-premises). But implicit in this need/ask for scale is that not only should I be able to scale but that it also needs to be lower cost. So even though customers are thinking from technology pov, the economic underpinning in terms of lower cost expectation is key driver for cloud adoption.
Now let’s take a look at how do the CSPs or vendors use and think of ‘scale’… it’s very well captured in this chart below (slide from Urs Hölzle’s keynote at Google Cloud Live, March 25, 2014):
This usage is in economic sense – economies of scale. So as CSPs get larger scale, their cost basis goes down, and price drops can follow through to drive customer adoption. So usage of the word scale converges for both customer and CSPs when seen through the economics lens – the premise being, that cost is major adoption driver for public cloud and as CSPs gain economies of scale that drives down their costs, they pass the same along to customers to accelerate adoption as well as capture the long tail of the market.
But as is apparent from Hölzle’s slide above, that’s not what the CSPs seem to be doing… why is that? There can be argument for preserving margins, but can a CSP actually upend this short term profit seeking by rapidly decreasing the cost to capture the platform and thereon monetize in the long term? This is where it starts to get interesting… let’s delve slightly deeper and explore the possible strategies that CSPs can adopt and what could the possible outcomes be.
Here below is my simplistic view of how scale is impacting the current cloud market. This graphic is showing the simple economies of scale curve being constituted of multiple short term Average Total Cost (ATC) curves:
Simple way of understanding above graphic is that short term ATCs represent a CSP expanding its capacity (i.e. building Datacenters laying fibers etc.)… and short term ATCs tend to go lower (until dis-economies of scale kick in longer term) due to factors like HW cost getting lower, CSP becoming more efficient etc. And there is a theoretical lowest cost point as short term ATCs cumulatively result in long term ATC curve (marked in red circle on the curve in graphic).
It’s clear that in the Public Cloud space only players with deep pockets will survive because getting to that scale where one has a viable and competitive cost basis is key to survival. And getting that scale would require huge amount of investment that only handful of companies in the world can afford, hence the big 3 – AMZN, MSFT and GOOG.
Also, seems that the pricing strategy at this time is to maintain a profit margin on the short term ATC curve. That’s essentially what Hölzle’s graphic is saying… and it is also implying that prices need to drop further and faster to follow cost decrease as a CSPs get rapid scale.
Now what strategic options does a CSP have in this market with just 3 players? This completely depends on what you believe the current dynamics of the market are and what the equilibrium would look like. To me it is clear that this is the time of land grab because the current state of the market is fight to get customers on to your platform. As in any other platform war we have seen (mobile: iOS and Android, Desktop: PC, MAC) that eventually in equilibrium state, the market devolves into duopoly and the leader is clearly dominant with other player being distant second. So if I was making the strategic decision, then I would instantly price down to the lowest point on the long term ATC (red circle)… of course the tough job of figuring out what that will be remains, but it will be moot as huge price war would ensue in the market with all players matching the prices instantly. Thereon, with lowest possible costs 3 things will happen: a) cloud adoption will shoot up exponentially, b) platform choice will revert to value differentiation (vs. current pricing driven one with AWS setting the agenda and others following) and c) player with shallowest pocket will have to bow out.
If you buy this above theory (btw it’s Game Theory in application as I try and play out the strategic choices game), then very interesting probable scenarios emerge – who will be the dominant platform, who will bow out, where will the monetization come from since the prices are essentially equal to cost, is this business even viable enough in long term to be in (another airlines like business)?