Practically given up for dead after a disastrous post-merger quarter a year ago, Cloudera has continued its comeback with its third straight quarter of better than expected results. As Stephanie Condon reported a few days back, non-GAAP Q4 net revenue came in at $211.7 million, beating analyst estimates by about $10 million and a penny per share. the company finished the year at $794.2 million.
More to the point, the quarter was not a lark, but the latest in a string of better than expected quarters following a disastrous Q1 last year where renewals tanked immediately following the Hortonworks merger. Losses of 38 cents a share, compared to Street expectations of 23 cents, sent then-CEO Tom Reilly packing. In retrospect, the company underestimated the disruption of the merger and the likelihood that customers would take a pause until they learned about the product direction of the combined company.
Cloudera had a narrow path to recovery. It needed to get the next generation product right and it had a narrow window in which to execute as Carl Icahn started circling in.
How it happened
To its credit, Cloudera laid out the path for the converged platform. But more importantly, Cloudera did not merely refactor the combined Cloudera and Hortonworks platforms, but seized the opportunity to reengineer it natively for the cloud and put the zoo animals in their cages were they belonged with fully integrated Shared Data Experience, Machine Learning, and DataFlow products.
Under the hood, Cloudera doubled down on execution. Like any enterprise software business, profitability has relied on a land and expand strategy reflecting that the upfront costs of customer acquisition for six-figure buys will be substantial, but that the margin comes as the customer renews and expands the footprint. To its credit, Cloudera made booking renewals more systematic. While the company is no longer reporting quarterly churn rates, after a year of upheaval, it has gone back to the pre-merger level.
The road ahead
With the new product portfolio, Cloudera has provided dual paths, including the Data Hub for lift and shift, with the rest of the portfolio reengineered into a cloud-native, containerized architecture run by Kubernetes. Big on data bro Andrew Brust provided the lowdown on the watershed release last fall.
Entering calendar year 2020, Cloudera has had both the classic Data Hub and reengineered Cloudera Data Platform in the market for a couple quarters. On the roadmap this year is a private cloud edition that will capitalize on the alliance with IBM Red Hat for an OpenShift-certified deployment of CDP. This reflects a broader trend across the bulk of the enterprise software industry to hedge their bets through cloud vendor-independent strategies. It responds, not only to the reality that they don't own their own clouds and don't want to mortgage their destiny to any single provider, but also for customers who likewise don't necessarily want to put all their eggs in AWS, Azure, or GCP baskets.
As we noted in our year ahead outlook, the 2020s will become the era of what we term "the Hybrid Default." Enterprises are looking to take advantage of the operational agility and efficiency that cloud-native deployment provides them, but they also want to keep their options open on whether to go to the public or private cloud. While the data gravity for much of the data that Cloudera Data Platform users are likely to work with originates outside the data center, for many sectors, such as financial services or life sciences, the data either already resides in-house or cannot easily go to a public cloud for regulatory or policy reasons. And that's where private or hybrid cloud comes in.
Cloudera's initial response for private cloud is through OpenShift, but in a call with analysts, chief product officer Arun Murthy didn't close the door to potentially supporting other Kubernetes distributions. Of course, private or hybrid cloud is another place where Cloudera may face another decision point down the road as the big three have also placed their claims through offerings such as AWS Outposts, Azure Arc, and GCP Anthos.
Another goal for the year ahead is opening self-service options. It's a natural corollary that, once you open a managed cloud service, that it makes sense to make the whole process as low-touch as possible. For Cloudera, it's also a way to get beyond the land and expand process that was necessary before this because of the long sales and go live cycles that were associated with what used to be called Hadoop clusters (note; this is the first time in this piece that we have mentioned the H word). That would certainly make the user experience smoother and deliver value faster. But beyond that, as a cloud managed service, we'd like to see Cloudera take the closed-loop approach to the next step by applying machine learning to optimize configuring and running CDP in the cloud – much as Oracle is doing with the autonomous database.
So how about Cloudera beyond Cloudera?
Cloudera has made an impressive recovery. It continues to face headwinds from the vast array of data analytic and machine learning services in the cloud. The ace up its sleeve is SDX, which provides a breadth of data governance that is missing from most of those other services. The underlying technologies are Apache open source projects, but the binaries that make it a unified product (rather than a bunch of zoo animals) are proprietary. So the APIs are open. In the spirit of, "if you can't beat 'em, join 'em," there's no technical reason why SDX could not be offered to customers of any or all of those services – EMR, HDInsight, Cloud Dataproc, or Databricks – that would otherwise be beyond Cloudera's reach.