Hadoop vendor Hortonworks filed for its initial public offering on Monday in a move that could turn out to be a good indicator of how big the market for the open-source big data software will be in the near term. The financial numbers Hortonworks released illustrate that selling Hadoop is a resource-intensive endeavor — and at this point still a relatively unstable one — so a positive response from public market investors will likely be because of Hadoop’s promise rather than its current profits.
The long and short of Hortonworks’ financial situation is that revenue more than doubled on a year-over-year basis during the first nine months of 2014, to $33.4 million from $15.9 million. If you use the company’s preferred accounting metric of “gross billings,” you’re looking at a jump to $53.2 million from $24.4 million. During that same period, the company’s operating loss increased to nearly $87.9 million up from $48.4 million.
[company]Hortonworks[/company] earns much of its revenue through technology partnerships and reseller deals with other technology vendors. The share of revenue attributable to one customer — Microsoft — has steadily decreased, but still represented 22.4 percent of Hortonworks’ total 2014 revenue through September. Revenue from its top three customers totaled 37.4 percent. The current Microsoft deal expires in July 2015, and all of the company’s partnership deals are non-exclusive — something evidenced recently as Microsoft touted a cloud-based partnership with Hortonworks’ biggest rival, Cloudera, and another big Hortonworks partner, [company]Teradata[/company], also entered into a deep partnership with Cloudera.
Still, Teradata maintains an 8.3 percent equity share in Hortonworks. HP invested $50 million in Hortonworks in July to give it a 5.9 percent share.
Those aren’t the kind of numbers that might typically get investors jumping for joy. In fact, they’re not much different (although smaller in sheer size) from the type of earnings that forced cloud-collaboration darling Box to postpone its IPO after first floating the idea in March. But the big difference is that Hadoop is a generally perceived as a high-growth market where specialists such as Hortonworks, Cloudera and MapR — which essentially created their own market out of open source software — are the dominant players despite the fact that Cloudera, having formed in 2008, is the oldest among them all.
Whether it’s from connected devices ranging from jewelry to farm equipment, or from the trillions of web transactions that happen everyday, companies will continue generating and storing much more data than ever before. As the Hadoop ecosystem of technologies continues evolving into a platform capable of storing all that data and hosting many types of data-processing and data-analysis applications on top of it, the companies building the foundational pieces will continue to grow.
Dan Primack at Forbes suggested we shouldn’t be surprised to see Hortonworks go public before the year’s end, and I’m inclined to believe he’s right. I also suspect the public markets will treat Hortonworks well. Despite the high level of mud-slinging in the Hadoop space, I’ve encountered little but respect for Hortonworks CEO Rob Bearden. And Bearden has assured me in numerous conversations (including in the Structure Data 2014 session embedded below) that the company’s finances are very much in order and poised to skyrocket in the years to come.
Regardless how it performs, the IPO should result in some interesting moves by [company]Cloudera[/company] and [company]MapR[/company], especially, in early 2015. They’re correct to make the obligatory “it validates our market” and “a rising tide floats all boats” statements, but a successful IPO will have them considering how to ensure Hortonworks doesn’t accumulate too much capital or too much mindshare. A poor showing by Hortonworks might have them second-guessing their own IPO plans — or perhaps moving a little faster in order to prove their companies are different.
Indeed, all three companies are rather different. Cloudera and MapR are likely burning through a lot of cash, as well, but their financials probably look quite different because they generate much, if not most, of their revenue from software licenses rather than support and services. And earlier this year, Cloudera, easily the highest-valued among them, closed a $900 million financing round that included roughly $740 million from Intel and gave the chip giant an 18 percent stake in the company.
Industry watchers have been watching these companies amass venture capital and shift their strategies for years, always waiting for the other shoe to drop and give a sign of how the Hadoop market will shake out. With Hortonworks now on the clock to become a publicly traded company, it looks like they’ll finally get their wish.
For anyone wants even more fodder for predicting how the Hadoop space might shape up, listen to the two Structure Show podcasts embedded below. The first is with Bearden from July, and the second is with Cloudera CEO Tom Reilly from April.