Small Blue and the Bonsai Datacenter


Bonsai by Andreas D., on Flickr
Creative Commons Attribution 2.0 Generic License  Via Flickr

IBM today announced new datacenter investment plans to bolster its cloud computing presence. They’re going to spend $1.2 billion to build 15 new datacenters (seen above, lower shelf). After some consultation with the Twitterati on matters of long division, it appears IBM is going to spend a whopping $80 million per datacenter. That may sound impressive until you consider that the big boys in cloud can spend half a billion or more per datacenter. Google’s most recently reported quarterly capex was $2.29 billion.

Perhaps IBM is has some special sauce that lets them go toe-to-toe with the big boys on the cheap? If so, they haven’t bothered to mention it and they’re not known for their low costs or frugality. I for one am disappointed IBM has stifled its usual impulse to pitch the mainframe as the obvious choice for the workload de jour. Surely there is a story to weave about cloud computing bringing the industry back to its timesharing roots, blah, blah, blah, mainframe uber alles? Or maybe they’ve beaten their heads against that wall enough to knock some sense into them.

In the absence of special sauce, it seems more likely that IBM is either confused about what it takes to play in the big leagues and/or lacks the financial resources. They continue to confuse cloud computing with web hosting. Do they really believe their Amazon depositioning is relevant or is it just an attempt to muddy the water? What does IBM say when Amazon utters the letters C, I and A? IBM also has real constraints on their ability to invest due to their prioritization of financial engineering over engineering engineering.

I will offer IBM some free consulting for their next big initiative to help them come up with some differentiation and a storyline beyond how much money they are going to spend. What are the odds that they find themselves budgeting a billion dollars for almost every initiative? (The way we knew they had given up on AS/400 was when their grand revitalization initiative was only backed by $125 million). “One billion dollars” has been the only page in their marketing playbook for a long time. Had I more time and dedication to the cause, I would collect all the “one billion dollar” announcements and assess their subsequent market impact. Because this is capex (and because it isn’t a round billion dollars), the $1.2 billion number is probably a real number unlike most IBM investment numbers. But there is at least one real billion dollar number for IBM and that is their Q3 revenue miss.

Various evergreen belittlements aside, IBM seems to have woken up to the reality of cloud computing and the existential threat it poses. The “it is early days for cloud” speaking point seems to have been retired. They’re overreaching and flailing around with announcements and advertising. but are at least trying to get into the discussion. But they still face an extremely difficult road. IBM’s ability to develop technology (the engineering engineering thing) has atrophied (“SmartCloud? Just kidding…”) and letting others do the technology development is risky (e.g. taking an OpenStack dependency in the absence of controlling your own destiny is looking a lot riskier). And IBM is operationally unproven across multiple datacenters. It is easy to needle AWS for outages, but another to avoid outages yourself. The real question is would anyone notice an IBM outage. Finally, IBM is constrained financially relative to the competition.

Many financial observers assume that because IBM is one of the few technology companies to have survived multiple generational technology transitions, they will successfully traverse this one as well. Past performance is definitely no guarantee of future survival in technology and IBM’s past transitions are notable exceptions to the broader industry history. And this transition is different in that the workloads in play are core workloads for IBM. With the minicomputer and PC transitions (the later of which was near fatal to IBM), the workloads in question were mostly net new and didn’t directly replace mainframe workloads. The cloud is taking core workloads, so even if IBM executes well and moves existing customers to its cloud, they will take a revenue and margin hit.

IBM is damned if they do, damned if they don’t. If they accelerate the move to cloud, they will undercut their existing business and miss their sacred financial roadmap. If they don’t, everyone else will partition up their existing business. Maybe they can thread that needle, but IBM has not shown any reason to believe they can successfully catch up to and compete with the leaders in cloud computing. Bonsai datacenters show IBM wants (or needs) to compete on the cheap.

2 responses

  1. You can substitute the word “Cisco” for “IBM” in most of that. Late to the Cloud party and underinvesting compared to the major players. Cisco have the added bonus of SDN looming.

  2. Yep – Cisco definitely in the same boat. The alternatives just not as mature yet.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Get Updates By Email