Ever since Tom Reilly joined Cloudera as chief executive in June 2013, he has faced speculation that he was brought in to spruce up the pioneering distributor of Hadoop cloud analytics technologies for a sale.
Those doubts have persisted, despite Reilly’s ambitious goals, thanks partly to the recent travails of Hortonworks, a direct rival whose shares plummeted after it announced plans for a $100 million secondary stock offering in January, and Tableau Software, whose shares plunged by half in February when the company reduced its revenue outlook.
But in an interview on Mar. 31 near the close of Strata + Hadoop World, the conference in San Jose it runs with O’Reilly Media Inc., Reilly insisted he has no intention to sell Cloudera. Instead, he’s looking to an initial public offering–though not in this volatile market.
Although Cloudera isn’t yet profitable, Reilly reckons it has plenty of cash to get there, and he thinks cloud software is finally proving itself with use cases that corporate boardrooms can understand–and pay for. Here’s an edited version of the conversation:
Q: There are a lot of companies with overlapping capabilities here at this show, but it seems like there’s a bit of a struggle for many of them to make money. Does that cloud of companies, as it were, make the opportunity tougher for Cloudera?
A: I have a different view of that. There are 100-plus companies here, and every one of them is very well-funded. The funding’s going because this is one of the largest, fastest-growing areas of spend in enterprise software. There’s a large land grab going on to capitalize on this opportunity.
It’s just the past five to seven years the world has become hyper-connected. It’s that interconnected world that is creating tremendous opportunity, or threats, for traditional enterprises. So if they want to re-engineer their business for the interconnected world, that means they have to get their arms around all the new idea sets. If you’re an auto manufacturer and you’re not building a connected car, you’re in trouble. So auto manufacturers are becoming data management companies.
Q: But a lot of the revenues so far have gone to the big guys such as IBM, Oracle, Microsoft. The big seem to be getting bigger, so what’s the role for the smaller companies here, and the early innovators like Cloudera?
A: We in the industry eat our own. We’re here to compete with the IBMs and the large players. But both the big guys and the startups are embracing Hadoop in this new era of computing. The embracing by the large guys is expanding the market by validating it. And it creates the opportunities for the next tier of players, like Cloudera.
Q: Where do you see your biggest opportunity for Cloudera and generally where are the biggest pools of value to be created around Hadoop and open source enterprise generally? Are apps on top of Hadoop the biggest opportunity?
A: Back in 2014, we spent a lot of time talking about animals in the zoo–Zookeeper, Pig, Impala–the technologies. In 2015, the conversation advanced and we talked about the collection of animals as a thing–a data hub, or what some call the data lake–to indicate to CIOs how all this fits into your data lake landscape.
What we’re seeing now in 2016 is the emergence of the boardroom use cases. They come down to customer insight, Internet of Things data-driven products and services, lowering business risk. And we’re mapping those to particular industries. You’re not going to turn to legacy solutions to solve these problems, because they involve new data sets. Hadoop is designed for that data set.
Q: How far along are you or your partners in developing the applications on top of Hadoop?
A: I don’t think we’ll see packaged applications like we did with the last set of enterprise technologies. Take usage-based insurance: The algorithms that determine what your product will be when you say “pay when you drive” will be the competitive advantage. So the last thing they want to do is buy packaged algorithms. They’re going to want to have their data scientists create their own and use that as a competitive advantage.
What we do is package as much of the building blocks as we can but then give them the tools to determine their specific offerings. We work with partners who are experts in each vertical industry–healthcare, transportation, financial services.
Q: What about direct sales or working with big integrators such as Accenture?
A: We have 2,200 partners–infrastructure providers such as hardware and cloud providers, tools providers such as data management, analytics partners such as Tableau, Qlik, SAS, and consulting groups such as Accenture. What our sales force does is assemble a team of partners for almost every opportunity. When a partner is attached to one of our projects, it tends to be larger and more successful faster.
Q: How big is Cloudera’s sales force?
A: We don’t give out those numbers, but we have surpassed 1,200 employees. There are more than 20,000 enterprises using our software globally. We have proprietary software on top of the core open source, and we license that software to more than 825 large enterprises. About 75 percent to 80 percent of our revenues are software licenses, 15 percent 25 percent professional services, and 5 percent to 10 percent training.
Q: There’s a lot of services businesses in open source.
A: Yes, but services businesses are not high-margin. That’s why we sell software. Investors want to invest in software companies, not services companies.
Q: Which raises the question of how investors are looking at your company. What does Fidelity’s recent 38 percent reduction of the estimated value of its Cloudera investment say about Cloudera or the broader industry?
A: It says nothing about the state of the company, because they have no data about our performance. The macro markets have changed and they’re adjusting their portfolios down. Every public company’s valuation is down. The fact that they elect to change their estimated valuation of a private company is just conservatism on their part.
Q: Does that conservatism have any impact psychologically on customers?
Q: Because a lot of people also look at Hortonworks and Tableau shares getting hit recently. Are customers wary as a result?
A: Let’s take Fidelity. Even after it marked down its investment, it’s still significantly up from what it paid. They’re very happy.
We’re not focused on it. We’re not a public company. We’re fully funded through profitability.
Q: But not profitable yet?
A: We have more than enough cash to become profitable. We’re not dependent on the public markets for financing. In three-four-five years, we’re going to be immensely valuable. I’m looking at our growth rates, I’m looking at the customers we have, I look at this event, the momentum behind this industry, and we’re the market leader.
Q: There’s some perception that you were brought in to sell Cloudera at some point.
A: No. I hope this is my last job. And I hope to work a very long time.
In fact, the last company I ran was a public company [ArcSight], that was acquired by Hewlett-Packard. I learned something in that process. I didn’t want to sell that company. Were I running it today, it would be an amazingly large company. But the minute you’re a public company, there’s a for-sale sign out. If someone wants to come along and pay you a premium … and you say no, your shareholders are going to be very upset. Private companies can say no.
We’re building a lasting, enduring enterprise software company.
Q: Obviously there are advantages to going public–access to capital, a presence. Red Hat has gotten attention not only from going public but getting to $2 billion in annual revenues. Is that ultimately your goal?
A: Yes. We fully intend to be a public company. We’re just fortunate that we can do it on our timing, when we’ve reached the right scale, when the business is more predictable, when there’s greater visibility.
We are of a size and scale today to be a successful public company. We’re operating well enough to be a well-run public company. It’s not a market that we’d want to enter as a public company right now.
Q: So that big slug of funding in 2014 [$900 million, including $740 million from Intel] really helped?
A: The Intel partnership has really improved our company and our operations. We have a good three- to five-year visibility into what’s coming in the silicon and they want our software to be optimized and designed to take future advantage of their silicon. Their hardware engineers interview our data scientists. They say, if you’re doing an anti-money laundering use case or a next-best-offer use case, how do you write your algorithms?
It turns out data scientists use a lot of subroutines consistently, so their hardware engineers go, if we can write an instruction set in the chip that can automate that part of it, we’re accelerating the performance. It’s Intel’s goal that if you design hardware and software together, you can get upwards of an 8X multiplier on Moore’s Law.
Q: How do you navigate the uncertainties of customers over whether to use public cloud versus private clouds versus hybrid clouds?
A: Public cloud is our fastest-growing environment. Our overall business is roughly doubling. Our cloud business is double that [growth rate]. It may be 15 percent to 20 percent of our customer workloads today. Most of our customers are operating in a hybrid environment.
Those new data sets–mobile, social data–that is being generated outside the data center, in the cloud. You want to marry it with data in your own data center, like customer contracts, historical records, customer support cases.
Data has mass. Data that’s created in your data center wants to stay there. It’s hard for banks or healthcare companies to put it in the cloud. But data that originated in the cloud also has mass and wants to stay in the cloud. All the new data is in the cloud.