BackIn big data, shepherding comes first
Building big data businesses is proving to be anything but a get-rich-quick game, and requires both agility and patience
Premium
Big data is increasingly moving into the mainstream, as companies in every industry begin to combine an abundance of digital data with smart software to analyse it. It is a potential gold mine for software-makers, with analysts predicting torrid growth for the market overall.
But building big data businesses is proving to be anything but a get-rich-quick game, and requires both agility and patience.
Take Cask, a start-up in Silicon Valley founded in 2011, backed by leading venture capitalists and led by former Facebook and Yahoo engineers. In late September, the promising young company changed both its name and its business model—moving to supplying open-source software and trying to make money on technical support and consulting rather than on proprietary products. “Everybody is scrambling to take revenue where they can get it," said Jonathan Gray, chief executive of Cask, which was previously Continuuity.
The challenge is similar for major tech firms pursuing the big data market, but it is most acute for start-ups, which lack the financial ballast of the tech giants.
The goal for every start-up is to quickly capture a foothold in the emerging market, then to grow rapidly and profitably. Most want to be software companies, because software is the most lucrative part of the technology business. And it scales up the fastest, since companies are selling code, typically delivered remotely over the Internet as cloud software, or technical support subscriptions for freely distributed open-source software—instead of selling computers or people’s time as consultants.
Yet data science, as a business, is still young. As the technology moves beyond the Internet incubators like Google and Facebook, it has to be applied company by company, in one industry after another. At this stage, there is a lot of hand craftsmanship rather than software automation. So the aspiring software companies find themselves training, advising and building pilot projects for their commercial customers. They are acting far more as services companies than they hope to be eventually.
For now, the technical hand-holding and pump-priming is an essential step. The young firms are building the market of commercial users in retailing, finance, health care and other industries to which they will then sell software.
“They are setting up and doing first deployments to get corporate customers up to speed," said Merv Adrian, an analyst at Gartner.
For the start-ups that survive an inevitable winnowing of suppliers, the rewards could be rich. The total market for big data technology, according to the research firm IDC, will reach $41.5 billion by 2018, more than tripling in five years. And there are far more optimistic forecasts.
Yet the payday will come only for those who find sustainable business models. The cost of chasing the big data opportunity can be daunting. A glimpse into one company’s bet was revealed last month, when Hortonworks filed documents to prepare for selling shares to the public. Hortonworks is a leading distributor of the open-source Hadoop software, which is a database for handling so-called unstructured data from the Web, sensors and smartphones, used in big data analytics.
The company’s financial statements show rapid growth in revenue, more than doubling in the nine months that ended on 30 September, to $33.4 million. But its costs surged, resulting in a net loss of $86.7 million, well more than double its total revenue. If Hortonworks proves a winner, the losses will be a bold investment in the future. If not, the losses will just be money lost. Optimism prevailed on Friday, the first day of trading in Hortonworks shares. Investors bid up the company’s stock to $26.38 a share, 65% above the offering price of $16 a share.
The big operating loss at Hortonworks, said Mike Gualtieri, an analyst at Forrester Research, “shows that it is still early for this market, and that you need to spend a lot."
Other big data start-ups with ambitious plans have raised substantial money from venture and corporate investors, but are choosing to remain private for a while. That way, they say, they can ride out market shifts, grow further and develop durable businesses before they have to report quarterly results and face pressure from public investors.
Cloudera, another distributor of Hadoop, raised $900 million this year, with $740 million of the total coming from the chipmaker Intel. Cloudera, analysts estimate, is about twice as large as Hortonworks. Mike Olson, the chief strategy officer at Cloudera, said the big fundraising gave the company both a strategic partnership with Intel and breathing room. “It affords us time to achieve growth, stability and discipline so we can thrive and prepare for the scrutiny of the public markets," Olson said.
Palantir Technologies, founded a decade ago, has raised a total of about $900 million to help develop its hybrid model of data analytics software and services. Founded in 2004, Palantir began doing work for US intelligence agencies, but its clients now include banks, hedge funds, insurance companies, pharmaceutical makers and public health organizations. The company has steadily built and refined its underlying layer of big data software, but nearly all its work is custom projects with clients involving a lot of consulting and advising. Palantir expects to generate revenue of about $1 billion this year.
Today, corporations frequently need help beyond data-handling know-how. Silicon Valley Data Science, a consulting firm, was founded last year by a group of data experts. Initially, recalled Edd Dumbill, vice-president for strategy, they thought their role would be to bring data skills to corporate projects that were already underway.
But often, Dumbill said, companies knew they had a problem, knew they had data, but not how to devise projects to explore and experiment with data. “So we had to move up to a higher level with clients to work on data strategy, identifying a road map," he said.
At Cask, Gray and his team are creating software tools to make it easier for corporate developers to write big data programs that run on Hadoop. The Internet giants, said Gray, a former engineer at Facebook, achieve what they do using “heroics," the labour of elite teams of software wizards. That is not a formula for bringing data science into the corporate mainstream. Instead, the programmers that work in banks, retailers, health care providers, media companies and elsewhere will be critical. “The industry experts will be the ones building these new applications," he said.
Until September, Cask had sold its software tools as a proprietary subscription service delivered over the cloud, which was a way to generate some revenue early on. But to gain wider adoption, the company made its current software open source—and changed its name.
Later, Gray said, Cask, like many open-source companies, may charge for added features, like management and security software. But today, revenue is coming from helping corporate customers start writing big data applications. Cask, he said, works with corporate developers, often building the first half of a pilot project and handing off the second half of the project to them. “It’s a bootstrapping and training role," Gray said. “We’ve all got to prime the big data pump. It’s early days."
©2014/The New York Times
Unlock a world of Benefits! From insightful newsletters to real-time stock tracking, breaking news and a personalized newsfeed – it's all here, just a click away! Login Now!
