Year end is always time for either year in review style articles or predictions for the upcoming year. Instead, as the clock inexorably marches on, I wanted to give a shout out to some old school technology. While it’s my job as a venture capitalist to spend time thinking about new disruptive technologies, it is fascinating to think about the venerable older technologies that are still enterprise workhorses and the tremendous amount of inertia start-ups must overcome to displace them.
I got thinking about this because I was recently on a panel where someone pointed out that a large portion of server revenue is in alternative Instruction Set Architectures (ISAs) i.e. anything non-x86 such as Power, IA-64/Itamium, SPARC, etc. This seemed like an interesting factoid so I looked into the data, which is indeed shocking. Although non-x86 serves, which I categorized as Everything Else in the chart below account for 1.8 percent of the volume of units, they are over 28 percent of the revenue in 2012!
In fairness, much part of this is due to the much higher average selling price of Everything Else servers like mainframes and SPARC, and of course in those business lines, the percentage of both units and revenue is decrease rapidly, but why are people still buying those machines? No one sets out today to build an application on Itanium yet $ 3 billion was spent last year on the systems.
But the answer isn’t that surprising when you understand big companies. Someone built an application that works fine and is sitting in the corner. The guy who wrote the application left the company seven years ago, the documentation sucks, and people are afraid to touch it. So the care and feeding of these applications drives this revenue.
But I just don’t want to pick on servers and chips vendors here. There are plenty of anecdotal and hopefully humorous data points from up and down the stack I can offer:
- A major financial institution that still has a few VAX machines running in the basement. That’s a 35-year-old architecture.
- A major media company that still has servers running Windows NT 3.5 This version of the operating system was released in 1994.
- A global financial firm running Internet Explorer 6. Yup, that’s circa 2001 technology.
These companies are dramatic example to prove a point about inertia. Over a beer, ask anyone working in technology at a large company and you are likely to hear the same types of stories. Inertia in the enterprise is high, otherwise these technologies would be long gone. Further, in an environment with flat-to-moderately-growing IT budgets, it’s a zero sum game. Budget for your new widget is taking food off someone else’s table. There are no overnight successes in the enterprise. Splunk, which went public in April, was founded in 2003; Palo Alto Networks, which went public in July, was formed in 2007; and even Nicira, which burst into the public’s mind in 2012 after selling to VMware for $ 1.26 billion, had been working away since 2007.
Sometimes the money isn’t in the bleeding edge. When you step out of the Silicon Valley echo chamber and the industry conferences with all the smartest speakers you quickly realize that most enterprises don’t see technology as providing them with competitive advantage. This is a debatable point in the long run but in the short term they have a business problem they are trying to solve and want to do so for the least amount of money possible. They are short on IT staff and more than willing to let the big vendors tell them what they need. Your OpenStack-Bigdata-NoSQL-Cloud-Openflow-distributed-real-time-anltyics-quantum-flux capacitor may be cool, and buzz word compliant but may fall on deaf ears in the mass markets.
But don’t despair, there are leading edge end-users out there who are willing to take a chance as early adopters of compelling new start-up technology. Plus, these are not just the big web companies and Wall Street banks. Sometimes you will find them in the most unusual places and it’s my New year’s resolution to spend more time with these folks in the coming year.