Tech sugar daddies shovel millions into Hadoop war
Who will flash the most cloud cash?
Open ... And Shut There was once an idyllic time when people like Joe Kraus described an entrepreneur's dream of starting robust companies on a shoestring budget, powered by open-source software and cloud infrastructure. Apparently Cloudera and Hortonworks didn't get the memo. Both Hadoop competitors recently raised mountains of cash at sky-high valuations, fuelled by open-source software and cloud infrastructure. And now Cloudera investor Ping Li has declared that his firm, Accel, is prepared to dump $100m more into Hadoop's meta-market, Big Data.
Well, venture capitalists do, for one, and at valuations that entrepreneurs might be unwise to pass up. In a hot market like Big Data, where Cloudera and Hortonworks compete, it seems that VCs are trying to preempt competition by going very big, very fast, in a scorched earth policy of sorts for would-be competitors. It's hard to imagine other VCs having the appetite to find other Hadoop companies when Cloudera and Hortonworks are so richly resourced.
Betting big on winners
TechCrunch has another explanation for the massive Series A valuations, and the amount of cash going into a select group of companies. When a VC has found a great entrepreneur and a market with proven traction:
[T]hey bid early and aggressively (hence the run up on these valuations). At the end of the day, to the venture firm, the most important thing is ownership in the deal. If they have to pay a premium on the valuation, it’s not a big deal because they have a large fund that can absorb the extra money paid for the “hot” company. The net net for seed stage deals however is that there is even more capital funding more companies and the takeaway for Series A deals is higher valuations (which require more investment for the target 20 per cent ownership for the venture firm).
This reasoning jibes well with feedback from Mike Olson, CEO of Cloudera. In an email exchange, Olson challenged the argument that these big financings are about covering big costs, so much as betting big to achieve big returns:
[Y]ou think about this in terms of cost. I think about this in terms of return. I have been in the database industry for a quarter century. I've done [been deeply involved] in open source software for a long, long time. I have been watching this roulette wheel go around a whole bunch of times, now, and I am convinced that I have the pattern. I know where it's going to stop next.
You want to tell me how much money I should put on that spot?
For Olson, and almost certainly for Hortonworks (Hortonworks' COO Rob Bearden and Benchmark investor Peter Fenton hadn't replied by the time of publication), they're raising significant amounts of money because the opportunity is huge. A Big Data gold rush is underway.
This was evident at last week's Hadoop World, which 1,400 people attended, representing 580 companies, 27 countries, and 40 states within the US JP Morgan Chase, eBay, CBS Interactive, Oxford University, and a range of other big brands were on the agenda, talking about real-world deployments of Hadoop at massive scale. There is money to be made, and the time for timid investing is arguably over.
After all, neither company is really competing against other startups. EMC is the 800-pound gorilla in the industry, with a deeply entrenched salesforce in every account Hortonworks or Cloudera might even think about calling. In that sense, the tens of millions of dollars these companies are raising is hardly even table stakes to get in the game.
Burning a hole in their pockets?
Even so, the problem is that money tends to burn holes in entrepreneurs' pockets. Cloudera's investors didn't put $40m into its bank account to provide a pension plan for employees. Cloudera says it's using its latest Series D round of $40m to "further expand its marketing and sales operations, and support key strategic initiatives".
This isn't about inhibiting the entry of other Hadoop-focused startups, according to Olson. Quite the opposite:
I am far from wanting to lay waste to new entrants – we need a bunch of companies to get in here and get to work building products and services that make this platform even more valuable. Some of those will probably choose to compete with us. I think that's a mistake, but the investment in the platform that they make will be good for them and good for us and most of all good for customers. The majority of the ecosystem, though, will build applications and tools and infrastructure and services that complement what we do. That's critical to driving adoption broadly by business users.
Fair enough. But someone still has to spend the money. That $76m that Cloudera has raised will be used to scale up operations at a speed that the company may not be able to sustain, even if the market is ripe.
Or consider Hortonworks. The company currently lacks a saleable product, but raised a $20m Series A round from Yahoo! and Benchmark just to get started. It's hard to see how Hortonworks can spend that money efficiently at this stage. It doesn't need more sales heads because currently all it has to sell is support.
But support at this stage, without its own Hadoop distribution (which is now in the works), really means "professional services". And professional services, while a fine business for someone like Accenture, is not what Benchmark and Yahoo! spent $20m to build.
And so Hortonworks, Cloudera, and other startups with more venture funding than customer revenue are left to figure out how to wisely spend huge piles of cash. It may sound like a great problem to have, but it's often the inverse. Just ask Groupon, which has used oceans of cash to pay back insiders and investors and scale up a loss-making business that only gets worse the bigger it grows.
Or ask me. Back in 2000 I worked for Lineo, an embedded Linux startup that aspired to and filed for an IPO, despite having less than $1m in revenues. We were told to grow big, fast, and with the $60m investors happily gave us we grew head count 10X while revenues remained largely flat. Our abundance of venture dollars actively inhibited our ability and need to think clearly about how to build a business in embedded systems.
In 2002 we were essentially sold for scrap metal to Metrowerks, a division of Motorola. This, despite all that investor money. Or probably because of it.
No choice but to grow?
Perhaps Cloudera and Hortonworks will avoid Lineo's fate. Both are run by exceptionally talented and experienced executives. But these same executives will be under pressure to spend that money they've raised, and it's dangerous to slam your foot on the accelerator if you're still assembling the car. Entropy is hard to contain with a tank full of cash.
It's possible, of course, that these firms have little choice. The market is hot and in open source one winner tends to take all. Raising a mountain of cash is one way to declare an early winner and allow that company to then focus its resources on growing the requisite community.
But it's not clear that winning the funding war tends to crown a victor in open-source markets. Great code, not sales and marketing headcount, tends to win in open source. Eucalyptus, for example, won the private cloud funding war long ago but its victory in the market is by no means assured as OpenStack and others have entered the market, with OpenStack doing a particularly good job of marshalling community resources even as Eucalyptus has collected hefty financial resources.
So perhaps it's positive that Cloudera and Hortonworks have engaged in a somewhat pointless chest-thumping exercise over which has contributed the most to Hadoop. It's not a helpful contest, and one that overlooks rising competitors like MapR, but it perhaps shows that for all their fundraising prowess, the two companies recognise the real war will be fought with and for developers.
Whether their fat bank balances will corrupt that vision remains to be seen. As stated, both have the benefit of great management. They're most definitely going to need it.
Matt Asay is senior vice president of business development at Strobe, a startup that offers an open source framework for building mobile apps. He was formerly chief operating officer of Ubuntu commercial operation Canonical. With more than a decade spent in open source, Asay served as Alfresco's general manager for the Americas and vice president of business development, and he helped put Novell on its open source track. Asay is an emeritus board member of the Open Source Initiative (OSI). His column, Open...and Shut, appears twice a week on The Register.