Bullish as I may be, however, I’m virtually a naysayer relative to Hugh MacLeod (AKA gapingovid), whose recent piece “The Cloud’s Best-kept Secret” made the rounds last week to near universal acclaim. One of my colleagues even joined the chorus of voices that seconded his basic contention: that from the current crop of cloud players could emerge a single dominant player, and that this player could be enormous.
Where enormous means: could be “the largest company to have ever existed.” Now that is bullish.
But while I agree with Tim that this is an important post, and with the 451 Group’s Rachel Chalmers that it’s potentially chilling, and I respect Hugh quite a bit, I can’t agree. Couldn’t disagree more, in fact. Even after giving myself a few days to ponder it, I don’t buy either contention.
Let’s take the assertions in sequence.
First, the idea that one firm may come to dominate the cloud, “the way Google came to dominate Search, the way Microsoft came to dominate Software.” Given the precedent, this contention, on the surface, is not all that far-fetched. Try as we might to avoid them, monopolies regularly recur in the technology industry. Probably because of the tendency for enterprise buyers to be both conservative and heads-down, myopically focused on the present. Those that forget history being doomed to repeat it, and so on.
Still, I see a couple of basic problems. One, there is the question of operating systems. If you believe, as I do, that operating systems still matter for the overwhelming majority of workloads, then a single cloud provider is unlikely. Linux, to date, has been the dominant platform of the cloud, but expect that to change. Microsoft will not willingly cede such a crucial market to a rival platform, hence the pending arrival of Red Dog and a few other Microsoft cloud initiatives. So we’ll have at least Linux and Windows in the cloud, and el Reg says that Solaris is on the way. It seems just as unlikely to me that one provider will adequately service the needs of multiple platforms as it does that customers will unanimously migrate all server workloads to a single platform.
Some might argue that virtualization is the means by which a single provider could achieve the dominance necessary, but I doubt it. Would you trust Microsoft with your Linux cloud? Or Amazon with your Windows cloud? Maybe you would, I probably wouldn’t: not on anything more than a tactical level.
But let’s assume for the sake of argument that you don’t believe the operating system is important; nobody mentions, for instance, the operating system when they talk about Google App Engine. They do, however, mention the choice of language – Python, in that case. However much you may be able to abstract the underlying operating system – even to the point of irrelevance – there must remain some developer target. Language, libraries, and so on. So choice remains, for both the cloud provider – what OS, language, database, filesystem, and son on – and the cloud customer.
Which means, as far as I understand them, that the Power Laws cited by Hugh are either not terribly relevant or trumped by the switching costs. Or both. Shirky’s piece, after all, discusses their applicability to “systems where many people are free to choose between many options.” Which does not describe any current cloud computing platform that I’m aware of, nor – as discussed above – those we’re likely to see in the foreseeable future.
If my application portfolio is Windows based, for example, I am not free to choose Amazon. If my workloads run on Java, I am not free to choose Google App Engine. If Python, then not Force.com. And so on.
Some of these restrictions, undoubtedly, will subside as platforms are updated, evolve and add new features. But not all of them. Operating system and programming language choices will remain, and therefore people will be counterintuitively less free to choose from the available platforms. And therefore less subject to the same Power Laws that Shirky described in conjunction with blog popularity curves.
Moreover, history tells us that court determined monopolies like Microsoft’s or even virtual monopolies like Google’s are more exceptions than the rule. The PC manufacturer market is not monopolized, the x86 server market is not monopolized, and – closer to home – the hosting market is anything but monopolized. Google and Microsoft’s market positions, in my view, are far more the product of unique and relatively rare market conditions (not to diminish the business acumen of either entity) than they are a technology industry inevitability. I regard both as anomalies, in this case, rather than the ultimate destination of industry trends.
The single cloud provider theory, then, I am skeptical of. Even should I be wrong on that score, I’m unclear on how, precisely, Hugh arrives at his predictions for the financial windfalls that await the would be King-of-the-Cloud.
I’d guess that he’s built that assertion on the possibility that the cloud becomes the ultimate realization of vendor ambitions: multiple inbound revenue streams from multiple lines of business – hardware, software, storage, customer telemetry (in the case of Google) – all within one customer. IBM and Sun, for example, could sell hardware, software, and storage to customers in one easy package – and likely to markets they’ve never reached before. Google, meanwhile, can profit at once from the application data generated and the fees charged to run said application.
That the cloud has revenue potential – or “synergisitic opportunities” as the finance guys might call them – I do not doubt. Just ask our customers what we’re telling them. Sometimes publicly.
But “the largest company to have ever existed?” I don’t see it, and simply can’t make that calculus work.
Of the Top 25 from the 2008 Fortune 1000 (size by revenue), but two firms could be considered technology representatives. HP and IBM. Neither are in the Top 10; HP checks in at #13 ($104,286,000,000), IBM #14 ($98,786,000,000). If you add their revenues together, in fact, they’re still only good for #4 on the list, behind Wal-Mart, Exxon Mobil, and Chevron. Add in the largest software firm on the planet (Microsoft #44 – $51,122,000,000) and the sometimes leader in shipped PCs (Dell #34 – $61,133,000,000) and you pass Chevron but still check in well behind ($314,327,000,000) Wal-Mart ($378,799,000,000) and Exxon Mobil ($372,824,000,000). Throw in the biggest chip designer (Intel #60 – $38,334,000,000) and you still don’t crack the top 2 ($353,661,000,000).
Put simply, if combining Dell, HP, IBM, Intel, and Microsoft isn’t enough to secure the largest firm by revenue title, I’m not sure how the cloud gets it there. I’m not against it – as I’m a believer in the potential of the cloud – but even I can’t defend that case.
The cloud’s important, but not likely to be that important. Or so say the numbers.