The Coming Cloud Crash

by: Dana Blankenhorn


Containerization is bringing the end of the cloud boom closer.

Leaders like Amazon and Google continue to spend furiously to bring a crash on.

The crash will take out many of the cloud industry's laggards and hasten consolidation.

The move from "virtualization" of cloud infrastructure to the use of "containers" for data and software on clouds is going to accelerate the next cloud crash.

And make no mistake, a cloud crash is coming. (AMZN) keeps pushing prices to the floor, while spending over $1 billion/quarter on new data centers. Google (GOOG) (GOOGL) is doing the same, and Microsoft (MSFT) is responding on the price front.

Other major players, like IBM (IBM), Hewlett-Packard (HPQ) and the phone companies - AT&T (T), Verizon (VZ), and CenturyLink (CTL) - are also having to invest capital in new data centers, with no assurance customers will be there to use them.

Containers are an important contribution to the cloud world. They bring everything an application needs, including all its data, software, dependencies and so on, into the cloud, enclosed in a software "container" of fixed size. This makes the whole thing more secure. It encourages large enterprises to move mission-critical applications, not just their customer-facing stuff, to the cloud. It's a good thing.

But, as with the AOL (NYSE:AOL)-Time Warner (NYSE:TWX) merger over 14 years ago, it really starts to put a period under demand for cloud. It also gives enterprises a tool they can use to gauge just how much internal cloud capacity they need, now and in the future, to gain maximum efficiency from the use of cloud technology.

In short, containers accelerate the move to the private cloud and put a ceiling on the growth of the public cloud.

The reason stock fell out of bed recently may be, in part, because some investors recognize this fact. App developer Huan Liu has been making this point for two years now, at his blog, talking about server utilization, which remains very low.

According to Liu's analysis, server utilization within the Amazon cloud - which dominates public cloud utilization - remains under 5% most of the time, except overnight, when batch operations and video streams can bring it up over 25%. This is something cloud providers have to account for, the possibility that spikes in usage may increase utilization for a short period of time, but that many servers will remain quiescent most of the time.

While it's dramatically cheaper for a cloud provider to build its own server racks, and surround them with infrastructure to maintain the lowest possible costs per server, than it is to buy-and-install a server rack, the difference is not infinite. There's a point where server utilization is so low that the savings from do-it-yourself cloud construction disappear, and where build-and-install starts to make sense.

Containerization brings that day closer.

For the biggest players, like and Google, their own internal needs can get them over this hurdle. That's why they're funneling capital into the field as fast as they can - to blow past the crash and raise prices once their competitors fall.

Their competitors will fall. I wouldn't want to be a manager at HP right now, nor at the phone companies, nor at a specialized provider like Rackspace (RAX) either. Because when the cloud "crash" happens, it's going to take out all these weaker players.

Disclosure: The author is long GOOG, GOOGL, AMZN. The author wrote this article themselves, and it expresses their own opinions. The author is not receiving compensation for it (other than from Seeking Alpha). The author has no business relationship with any company whose stock is mentioned in this article.