Cloudera Tumbles

Cloudera was a great place to work. I had lots of fun there, and Clouderans are the best people. So I’m genuinely sorry to see the company run into trouble.

Last week, Cloudera delivered its first post-merger earnings report. A bloodbath ensued:

Cloudera Inc. was swimming in downgrades on Thursday morning as analysts digested the company’s weak outlook as well as its disclosure that its chief executive would be stepping down, and shares plunged. At least four analysts cut their ratings on the stock following the first-quarter report, arguing that it will take time for the company to turn its business around and that things could get worse for Cloudera before they improve. The stock opened down 38.5% from Wednesday’s close.

Pundits greeted the news with predictable hot takes. “Cloud is eating the world!” say some analysts. “Hadoop is dead!” say others. These are half-truths, at best.

— Many Cloudera customers run their clusters in the public cloud or virtual private clouds. Shifting IT services to the cloud does not necessarily mean shifting data away from Cloudera. It does expose Cloudera to competition from “store brand” distributions, such as AWS EMR and Google Cloud Dataproc. (Microsoft uses HDP components in Azure HDInsight, and there is a business partnership between the two companies.)

— Data lakes are past peak hype, but not dead. Even Nick Heudecker, Hadoop’s biggest bear, thinks there is a role for data lakes in an enterprise data solution.

Industry analysts have declared the death of Hadoop since <checks notes> 2012.

Two charts explain investors’ dismay. The first chart, below, shows revenue reported separately by Cloudera and Hortonworks last year compared to post-merger revenue. Total revenue post-merger is just 3% greater than the combined revenue of the two companies in Q1 2018. And it’s significantly down from the combined revenue reported in Q3 2018.

That’s stunning.

It’s not “we’re struggling to grow as fast as we did two years ago.”

It’s “some customers pulled the plug.”

Source: SEC Form 10-Q. Hortonworks used a calendar fiscal year, while Cloudera uses a February through January fiscal year. Neither Cloudera nor Hortonworks filed 10-Q reports for Q4 2018.

Folks at Cloudera used to say that if they didn’t have to compete with Hortonworks’ free distribution, they could charge more for Cloudera. I guess that wasn’t true.

The second chart shows operating expenses for the two companies pre-merger and for the combined company.

Source: SEC Form 10-Q.

When Cloudera announced its acquisition of Hortonworks, I wrote this:

Cloudera and Hortonworks products are identical to everyone not employed by Cloudera and Hortonworks. ClouderaWorks will not need two sales forces, two marketing organizations, and two finance departments. It won’t need two MapReduce teams. The press release touts $160 million in cost savings. Those savings won’t come from pencils and duplicate magazine subscriptions.

Cloudera, this is Wall Street calling. Where are my cost savings?

From the moment that Cloudera announced this acquisition, the company needed to comb through the organization and identify redundant products and people. It’s not a fun job, but it needed to be done.

R&D is down from Q1 last year. Good job, Cloudera. Firing engineers and developers is just what you need to do if you want to stay competitive.

But G&A is up. Gotta keep an extra accountant around to compute those rosy non-GAAP numbers for the press release.

Sales and Marketing expense is up big time. Cloudera now spends sixty-four cents in sales and marketing costs for every dollar of revenue it earns.

Sixty-four fucking cents.

Hey, we just made a $100 sale. And it only cost us $64 in Cloudera t-shirts and coffee mugs to land that whopper.

They’d be better off firing the entire Marketing department.

To its credit, Cloudera has taken baby steps to clear out the weaker Hortonworks projects. Searches on the Cloudera website for Phoenix, Storm, Tez, and Zeppelin all turn up empty. But Cloudera still supports two distributions, two cluster management tools, and two security projects. And that’s where the going gets tough.

There are people at Cloudera who advocate tirelessly for their project. They can talk until they are blue in the face about why Apache Ambari is a better cluster management tool than Cloudera Manager.

Outside of Cloudera “Galactic” Headquarters, nobody gives a shit.

Existing Hortonworks customers don’t want to be forced to rip out Ambari and replace it with Cloudera Manager. I get it. They paid good money for that software, and they don’t want to see it deprecated.

Oh, wait. They didn’t pay for it. It was free.

When I wrote back on April 1 that Cloudera has no plans for a “fusion” distribution but will simply kill “all of that Hortonworks garbage” it was a parody. But was it really?

So, hard choices. And some people currently employed by Cloudera won’t be employed much longer. The comments on are priceless.


  • David Copperfield

    “R&D is down from Q1 last year. Good job, Cloudera. Firing engineers and developers is just what you need to do if you want to stay competitive.”

    Unless I’m missing something, it’s up 43% from Q1 last year. Or do you mean that it’s down as a fraction of total expenses? That’s true, but reducing the fraction of expenses in R&D while increasing the absolute amount wouldn’t require “firing engineers and developers”. Not sure what point you’re trying to make here.

  • Schabenbuggered

    I’m sorry to hear that Cloudera has fallen on hard times. My belief is that the description provided sounded much like SAS. I also believe that Cloudera’s failure may even embolden SAS management to keep making the dog food that dogs don’t like eating.

    Hypothetical, unsolicited advice to other analytics vendors – don’t replicate SAS practices in any way shape or form in your business. I believe that those practices won’t work in the real world, and won’t even work at SAS. Cloudera seemed to have something akin to a balance sheet for review. You’d never see that from SAS.

    I believe it would be fun to calculate the “Cost of Children’s Allowances (legitimate) per Dollar Revenue” and “Cost of Children’s Allowances (illegitimate) per Dollar Revenue” and see those on a graph. All hypothetically speaking, of course.

    Love and kisses, Schabenbuggered

    • One important difference between SAS and Cloudera is that SAS is a profitable company. Profitable companies prosper even when they don’t have a great growth story. Cloudera, on the other hand, has never been profitable — it’s valuation always depended on a growth story. Take away the growth story and you have toast.

      Prior to the merger, Cloudera was on track to break even. Hortonworks was not. Merging the two companies ruined them both.

      • Schabenbuggered

        I won’t argue SAS and profitability. They must have done something right at one point in order to finance all those castles and PR dollies. Annual proclamations about “revenue growth”, in my opinion, seem like tricky language intended to placate a flock unsophisticated with financial matters.

        Last I knew, [ Revenue ] – [ Expenses ] = [ Profit ]. Only one side of that equation is ever discussed (revenue). Excessive reserves (savings) can cover shortfalls, but that is not profit. Are they still “profitable”? Only a very small handful know the true answer. My belief is “No”.

      • You don’t need proclamations. If it’s a public company, check the 10-Qs. A company either has a growth record or it doesn’t.

        Investors are forward-looking. They will tolerate losses as long as there is revenue growth if they believe the company has positive operating leverage and can reach profitability through economies of scale.

  • Schabenbuggered

    The following statement was copied and pasted verbatim from a job description found online:

    “- Ability to work with and administer analytic and visualization tools, including Tableau and SAS Visual Analytics (this will be going away), and leverage them to build and design visualizations that transform data into actionable insight”

    Note the part in parentheses and the product immediately preceding it. How do you interpret a public statement such as this? Job description faux pas, loosely veiled outrage, or something else entirely? At the very least it provided a hearty chuckle.

  • From its inception, SAS positioned Visual Analytics as a “Tableau killer.”

    You will note that Tableau is not dead.

    Visual Analytics is not a bad product. It’s just not as good as Tableau. And any experienced Tableau user can see the difference after a short hands-on session.

  • Too quiet around here Dinsmore. We need you now more than ever.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.