Alex Cho
All posts from Alex Cho
Alex Cho in Alex Cho,

AWS versus IBM Let's Put This Baby to Rest

There has been some debate among readers here with regards to Deutsche Bank’s report, and whether AWS can sustain its lead against IBM. From my perspective, I think Deutsche Bank makes some fairly solid points in their report, as I have read the entire 33 pages from cover to cover. For the most part, IBM doesn’t have enough scale to compete with the major cloud providers, as has been broadly reported by many industry sources over the years.

However, the key area IBM plans to remain relevant inside of has more to do with its recent Soft Layer acquisition for $2billion. That business unit focused specifically on hybrid-cloud instances using Bare Metal (single-tenancy servers as opposed to multi-tenancy servers), this approach is somewhat differentiated from Amazon and Azure in that the public cloud servers of Amazon and Microsoft utilize a multi-tenancy server, which means that numerous server instances are running out of a single server, which means less usage of VMs (virtual machines). Therefore, the argument is that IBM can compete via its recent Soft Layer integration via its Bare Metal approach, but will struggle with gaining larger scale adoption as its dedicated to a shrinking niche. I.E. people don’t want their own dedicated servers in the cloud, especially as private cloud deployments seem to be taking off.

There were numerous points from Deutsche Bank’s report, but here’s one of them:

Several sources have told us that IBM SoftLayer has a relatively small-scale "pod" or mini data center architecture, with each pod having just 5,000-7,000 servers. Across 46 global data centers, this would equate to the server footprint of 275,000 noted above. It is hard to be precise, but we estimate that AWS and Microsoft have well over a million physical servers each, dwarfing the capacity of IBM SoftLayer.

So, IBM hasn’t invested too significantly into its server footprint, but even less convincing is the contrived reporting of cloud revenues to shareholders.

Here’s another salient point from Deutsche Bank with regards to IBM’s reporting of cloud revenue:

IBM has disclosed that $5.0 billion of its $10+ billion of total cloud revenues is generated from the sale of “foundational” offerings, whereby IBM sells the software, services and hardware to enable customers to “build their own clouds” (private data centers). This is evident in the split of “cloud” revenues in the prior Figures, which show that a large portion of “cloud” revenues are derived from hardware and professional services, in many cases to help enterprise customers build on-premise “private cloud” infrastructures. As a result, the $10.8 billion in trailing 12-month revenue that IBM is deriving from its cloud portfolio appears to be based on a relatively expansive definition of “cloud” that encompasses private and public cloud services. In contrast, AWS, Microsoft Azure and Google Cloud Platform revenues are derived from “pure” public cloud services (we define the difference later in this report).

Basically, IBM has been moving some of its more conventional revenue resources into the broader definition of cloud revenue, which negates the organic growth equivalent to peers within its respective space. It’s likely that IBM is losing share as a result of a broader transition to public cloud, and cannot sustain comparable margins to some of the hyper-scale providers, as such investors price this earnings premium into Amazon despite the lack of visibility on profitability on a quarter by quarter basis.