Red Hat is, without a doubt, the poster child for an open source subscription-based business model. This business model has come under scrutiny of late, stemming from a recent interview in which Red Hat’s CEO explains that reaching $5 billion in revenue requires replacing $50 billion in revenue currently flowing to other computer companies.
After joining Red Hat in December 2007, Whitehurst was quoted:
When I look at the quality of our existing technology, and the incredible brand that we have and the markets we play in, we should be a $5 billion company or more. If you just look at operating systems and middleware–that’s nearly a $100 billion business. We’re a $500 million business. We have barely scratched the surface.
Whitehurst and team grew Red Hat’s annual revenue to nearly $750 million in their most recent fiscal year. Computerworld UK columnist Glyn Moody asked Whitehurst whether Red Hat would reach the $5 billion revenue mark he had originally aspired to. Moody details Whitehurst’s response:
His answer was a good one. He said that he did think that Red Hat could get to $5 billion in due course, but that this entailed “replacing $50 billion of revenue” currently enjoyed by other computer companies. What he meant was that to attain that $5 billion of revenue Red Hat would have to displace software that currently costs $50 billion. Selling $50 billion-worth of software — even if it only costs $5 billion — is somewhat hard, which is why it will take a while to achieve.
Before I go on, let me explicitly state that I have a lot of respect for Red Hat and its people. In my role at IBM, I compete with JBoss, a division of Red Hat — but this post isn’t about JBoss. It’s about Red Hat’s business model choices.
I don’t accept Whitehurst’s explanation as to why reaching $5 billion will take longer than he may have originally anticipated.
Whitehurst seems to be saying that every $1 made by Red Hat results in $10 of existing revenue loss to established vendors. I won’t question the 1-to-10 ratio, as I have no data to offer.
I will, however, question the base assumption that Red Hat revenue comes at the direct loss of existing revenue enjoyed by “other computer companies,” as Whitehurst puts it. This may once have been true at the Linux level, where Red Hat Enterprise Linux (RHEL) has been a migration path for Unix customers.
But are masses of Unix customers still looking to migrate to RHEL? Or is it more likely that the majority of customers with Unix workloads today are going to remain on Unix? IDC’s estimates of server operating system market opportunity suggest as much — the revenue opportunity for Unix operating systems is forecasted to be flat from 2009 to 2013. If RHEL was growing largely at the expense of existing Unix revenue, one would expect a decline in these IDC figures.
I would assert that Red Hat’s Linux revenue growth potential is no longer linked to displacing another vendor’s revenue. Rather, the revenue potential is linked to growing usage of RHEL for new projects. Being the incumbent, Red Hat should find it easier to sell more RHEL licenses into an account than before, when it had to displace an alternative operating system to get into the account.
Yet, Red Hat’s business model, which focuses on low cost, has trained its customer base to fixate more on the price of Red Hat products than the value these products deliver. It encourages customers to trade RHEL for cheaper options.
Next, considering Red Hat’s JBoss business, it’s difficult to argue that JBoss is growing at the expense of IBM or Oracle. For instance, only 13 out of 133, or 9.8 percent, of JBoss-related customer case studies involved a migration from an establised enterprise application server to JBoss. Do migrations happen? Of course. But much more often, customers select JBoss for new projects alongside projects running on established enterprise application servers. However, by training customers to seek out low cost, rather than value, JBoss is now exposed to the likes of VMware SpringSource tc Server and other open-source-based application servers that are less expensive, if also less rich in features.
RHEL and JBoss products continue to perform well in the market and to drive revenue for Red Hat. That is without question. It will take Red Hat at least 13 years to reach $5 billion in revenue if the company is able to maintain the 15 percent year-on-year growth rate it achieved in the past year.
My issue is with Whitehurst suggesting that Red Hat’s revenue gains are linked to another vendor’s revenue slide. This may have been true in the past, but not any longer. Red Hat’s revenue growth is much more related to increasing workloads at existing customers or customers considering new projects. As Red Hat attempts this, the company has to realize that its focus on low cost has benefits and consequences.