Almost every industry produces huge amount of data, and the realization of the value of that data has driven the Big Data phenomenon. Powering this collection and analysis of vast quantities of data is a distributed system of clusters and nodes called Hadoop. Developed in 2005 by engineers at Yahoo!, this breakthrough in open source computing has spawned an entire ecosystem of vendors and investors, and for good reason: Allied Market Research estimates that the “Hadoop-as-a-Service” market will grow to $50.2 billion by 2020.
Two leading companies, Cloudera and Hortonworks, buoyed by hundreds of millions of dollars in VC funding, are fighting hard to get what Gartner estimates to be the 90% of enterprises without any modern Big Data solutions on board. Hortonworks filed for its IPO this week, and Cloudera is rumored to be not too far behind.
Cloudera’s first mover advantage
Cloudera became the first commercial vendor of Hadoop-based software when it was founded in 2008 by a team of former Facebook, Google, and Oracle engineers. The idea was, and continues to be, to provide an enterprise-ready distribution of Hadoop, plus the additional training and customer support. But its proprietary Enterprise Data Hub, announced earlier this year, has signalled that Cloudera is preparing to compete with not just its fellow Hadoop vendors, but companies like IBM on advanced data analytics and enterprise search.
It is estimated that the company has over 350 paying customers, including Monsanto, Allstate, and the U.S. Army. But just as important to Cloudera’s long-term success is the long list of corporate partners pushing its service. The company breaks them down into several categories: resellers (NetApp, HP), system integrators (Oracle), software (MongoDB, SAP), and hardware vendors (IBM, Dell).
Cloudera raised over $1 billion in March of this year alone, from Intel Capital, Google Ventures, and T. Rowe Price. With that much VC financing, an IPO appears to be inevitable. In fact, Cloudera CEO Tim Reilly, who took enterprise SI company Arcsight through its IPO and subsequent sale to HP, and who was hired in 2013 to replace co-founder Mike Olson, has basically confirmed that it will happen some time in 2015.
When it does, the valuation will have plenty of intrigue. In purchasing an 18% equity stake in the company, Intel Capital paid $780 million ($3 per share), to give the company an overall valuation of $4.22 billion. However, there’s a strong argument to be made that such a figure represents an overvaluation.
Hortonworks files for IPO first
Hortonworks set up shop in Palo Alto two years after Cloudera, but any head start Cloudera might have enjoyed is disappearing thanks to a different but compelling HaaS offering. Hortonworks, a Red Herring Top 100 winner, is reportedly attracting 60 new customers per quarter over the last year, including some big accounts like Spotify, eBay, Bloomberg, and Samsung. The company has also tabbed Microsoft and Teradata as corporate partners, creating powerful connections that are pushing the service as a complement to their own established data management user bases.
Hortonworks differentiates itself from Cloudera by employing the Red Hat strategy, whereby the company’s entire stack is comprised of Apache Foundation technology, none of it proprietary, with revenue instead coming from support and services. As the weight of the developer community continues to be behind an open source distribution of Hadoop, the bet is that the product likewise will continue to be good enough for most customers as well. If that’s the case, Hortonworks is your vendor.
Hortonworks has been supported through its early years by nearly $250 million in venture funding. Backers include Yahoo (19.6% equity stake), Benchmark Capital (18.7%), Index Ventures, Terradata, Hewlett-Packard, which financed Hortonworks’ latest round, a $50 million Series D raised this July, and Black Rock and Passport Capital, which led the company’s $100 million in equity financing before that. Hortonworks’ filing with the SEC revealed unsurprisingly that company is not yet profitable, with revenues ($33 million through the first nine months of 2014) growing at only a slightly quicker pace than losses ($86.7 million over the same time period).
So, what does it mean for HaaS now that Hortonworks has announced its intention of going public?
The IPO filing suggested that Hortonworks expects to raise $100 million from the public markets, but this appears to be just a placeholder number. While there is a PR boost associated with filing first, Cloudera and fellow Hadoop vendor MapR will have the benefit of calibrating their own offerings based on how much Hortonworks’ raise deviates from this initial estimate.
There is certainly room for three publicly-traded vendors in this fast-growing market. But there will be plenty of reshuffling in market share before Hadoop-as-a-Service reaches its steady state.