HDP - Saul?

Investment Analysis Clubs Saul’s Investing Discussions
2

Was wondering if you can talk about your position in HDP. I believe you recently bought back in.
I’m interested to know what motivated you to buy back in after selling.

Hi AJ,

It was Bert. I had quit in April, as you point out. I felt had good reasons, so I wrote to Bert about it, and he was very reassuring, helped me understand slightly better what they do, and pointed out again the enormous deferred revenue that they have in the bank ($198 million, at present). It’s subscription income and comes in at high margins. So I took a small position back. It’s currently just a 1.9% position, which puts it in 13th place (and my only really small position)… I’ll probably keep it this time, but beware, that’s “probably”!

Saul

3 Likes
3

Agh - another spoiler alert needed. There’s going to be fewer surprises to be had at the month end review this time.

Saul - I hadn’t picked up on your re-entry. Whilst the deferred revenues are definitely significant and growing it doesn’t really takeaway from the loss levels they face on an in-year basis which have increased as revenues increased - although I see they have held level to a degree in the last few quarters so maybe they have turned a corner.

Did you feel HDP was still the best way of playing the data integration space and not adding instead to Talend or Mulesoft? Do you seem them as an overalapping investment either in competition with each other or for your investment $s?

Thanks
Ant

4

what i read from magic quadrant chart is that hortonworks is either well balanced or does not know whom they are.

https://mapr.com/blog/gartner-2016-magic-quadrant-data-wareh…

2 Likes
5

Not sure that those guys listed in the magic quadrant are comparable!
Ant

6

March 2016. Anyone aware of 2017 report?

KC

7

Did you feel HDP was still the best way of playing the data integration space and not adding instead to Talend or Mulesoft? Do you seem them as an overalapping investment either in competition with each other or for your investment $s?

Hi Ant, I’ve increased Mule and Talend until they are both almost 6% each, while HDP is under 2%, which gives a pretty fair estimation of my evaluation.
Saul

8

Got it thanks mate.
Ant

9

Hi again, Ant,

Here’s what Bert wrote about HDP. It’s been about two weeks so he probably won’t mind that I post it.

Twilio has a model that almost made it inevitable for the company to have the results it recently reported. It charges for usage and users will simply not be willing to pay the enormous sums involved in what is really a messaging service. There are just too many alternatives.

As to Hortonworks, Hadoop is a methodology for storing data. It is far more efficient to store masses of data in Hadoop than in a traditional relational data base. I think the order of performance improvement is 100X-something like that. If a user wants to develop a big data application, then one choice is Hadoop. (There are other choices.)

In any event, Hadoop projects are hard to implement. Most users do not employ an army of data scientists and do not want to in any event. They want to use Hadoop which is open source, i.e. free, but cannot because it is far too difficult to use Hadoop (as it is) in a commercial environment. And so you have 3 companies-really that is all there are, who market services to make Hadoop usable. They are HDP, Cloudera which just went public and mapR. There are differences between these 3 vendors in that HDP is the only one of them to embrace “pure” Hadoop. (This would be a long discussion, not worth it in terms of investing).

Creating a Hadoop distro as it is called is shockingly expensive. That, of course, is why the companies exist-it is so expensive that no commercial enterprise can afford to create the tools needed to implement a Hadoop project. And that is the primary reason why the Hadoop companies lose money (the expense). If you looked at HDP earnings, one thing you notice is that more and more of their revenues are coming from subscription. In their case that means people are paying for a long term license to use their tools. It is a very healthy trend and is gradually raising gross margins. I think the rest of the income statement was pretty much as you may have anticipated. The company is on a path to profitability which is probably not a short-term project.

Because the shares are relatively cheap and remain so despite the appreciation, the odds are that HDP is going to get bought at some point. It would make a decent strategic acquisition for all of the companies in the DB market these days such as IBM, Oracle, MSFT and even SAP. They all need what HDP has. Oracle is the least likely given the character of it CTO/founder. In the mean time I expect steady progress towards profitability, nothing dramatic but more and more large wins and a shift toward subscription revenues raising gross margins.

I hope this helps. It’s obviously not a category crusher like Talend and Mulesoft, but it does have its positives, and revenue was up 51% last year, and deferred revenue was up 73%. I’ll keep a little position and see what happens.

Saul

11 Likes
10

It was Bert.

Saul,

Thanks for the explanation. I was just trying to get a sense for what changed your mind…not blindly following I can assure you. I have a bit higher conviction in the company. Operating leverage is beginning to show in the business. GAAP Op margins are “only” negative 97% of revenues which sounds horrific I must admit. However, GAAP losses one year ago were -158% of revs. On a non-GAAP basis they improved to -58% from -84%. Still a ways to go I’ll admit, but good to see the light at the end of the tunnel.

Their ownership/leadership is another compelling reason to own the company. They seem to have the cream of the crop in the Hadoop space. Bert has written that implementing Hadoop is an expensive process which is why these companies currently lose money. There are competitors in the space for sure, but not many. Cloudera recently IPO’d and is valued quite a bit higher than HDP. CLDR is almost at 10x sales right now whereas HDP less than 4x sales and 2.63x when you strip out cash (EV basis). If Cloudera truly has a better mousetrap, then the valuation may be reasonable. Or their valuation may simply come back to earth toward HDP. The flip side is that HDP is undervalued currently.

This will be an interesting space to follow for some time to come. Hadoop is billed at something like 100x improvement in performance, energy usage, etc…It is expensive and difficult to implement which is why these companies are cropping up.

In Hortonworks, I like the operating leverage beginning to show, their leadership and the current valuation.

Take care all,
A.J.

3 Likes
11

They seem to have the cream of the crop in the Hadoop space. Bert has written that implementing Hadoop is an expensive process which is why these companies currently lose money. There are competitors in the space for sure, but not many. Cloudera recently IPO’d and is valued quite a bit higher than HDP. CLDR is almost at 10x sales right now whereas HDP less than 4x sales and 2.63x when you strip out cash (EV basis). If Cloudera truly has a better mousetrap, then the valuation may be reasonable. Or their valuation may simply come back to earth toward HDP. The flip side is that HDP is undervalued currently.

AJ,

That is very much how I see it too, FWIW. It’s a balancing of many factors. Yes a PS of less than 4 seems cheap. But yes they’re losing money. However, they have deferred subscription revenue that they will receive for years without spending another dime. And probably some booked and not even billed yet, and/or likely to renew. I don’t really even get all in the weeds about it because it’s so hard to value.

I do think it’s probably partially that HDP is undervalued and partially that CLDR is overvalued, relatively speaking. Both could easily be targets for acquisition, as Bert says. And again as he says, it’s nice that Hortonworks is the most attractive, price-wise. Seems to put a floor under the shares, as I believe he said months ago in a Hortonworks article. I certainly see more upside than downside for HDP, which is why it’s one of my top positions.

Bear

2 Likes
12

That, of course, is why the companies exist-it is so expensive that no commercial enterprise can afford to create the tools needed to implement a Hadoop project. And that is the primary reason why the Hadoop companies lose money (the expense).

Bert is a great analyst, I appreciate his writings. However, I disagree with the above. If you really look into HDP’s income statement you would see the reason HDP is losing money is due to their high SGA, not because of R&D or labor costs. They are spending heavily on land grab. I am bit lazy but I believe I have written about this in detail sometime back.

4 Likes
13

Here is the comments I posted sometime back…

http://discussion.fool.com/every-question-you-have-for-hdp-is-va…

In this post I also shared my view of why Cloudera gets higher multiples whether it is justified is a different question.

3 Likes
14

Kingran,

In this case, I don’t get your distinction between “enterprise software” and “professional services.” Both companies are providing or attempting to provide the same product while one is more open source.

It doesn’t seem the distinction affects the cost structure.

Let me know what I’m missing.

Thanks,
AJ

15

Both companies are providing or attempting to provide the same product while one is more open source

The difference is Cloudera puts more tools, proprietary software like cloudera management suite, cloudera navigator, some machine learning stuff around Hadoop, that’s their differentiator, value add. They believe it adds more value to their customers, compared to open-source Hadoop and market thinks HDP, is just an open-source distributor, who primarily makes money from the installation, and run business. There is some merit to that argument. How much is what makes the market.

2 Likes
16

Here is the comments I posted sometime back…In this post I also shared my view of why Cloudera gets higher multiples whether it is justified is a different question.

Hi Kingman, that post of yours from Apr 30 that you linked to is really excellent.
Thanks
Saul

17

Kingman,
Thank you for linking to your old post, that is a fantastic insight!

This actually reminds me of Intel vs AMD. Intel has long been involved in setting international standards (IEEE, etc.). No short term benefit, but by being involved they can adjust the corporation direction early to stay aligned with upcoming open standards. This often helped them stay a step ahead of AMD and other competition.

If HDP is doing the same, that is an interesting investment in their own future which will never show up directly in their finance statements. This is a long-term focus I like in a company!

1 Like
18

Hi Kingman, I asked Bert about what you said (as follows):

HDP is a pure play Hadoop vs Cloudera who add their own tools, flavors, and I would say is about 80% Hadoop and 20% propitiatory. So why does it matter? Cloudera claims they are an “enterprise software” company, meaning they are investing in building software. In their growth phase they are investing, therefore they are incurring loss, hence their valuation, i.e., almost 9x revenue vs 3x for HDP. On the other hand, a pure-play hadoop provider who primarily makes their revenue through “professional services”, support and subscription should not be incurring such heavy losses.

However, HDP, is involved in Hadoop initiatives much deeper than anyone else, and they incur significant cost on that, but that is not recognized as “enterprise software” building because most of their work goes towards “open source”. While HDP model is slightly different from “enterprise software” companies, I think HDP margin’s can change dramatically. In the last Qtr HDP incurred $57m loss vs $54M revenue or 100% of revenue.

I think there is a land grab happening in Big data, “hadoop” world. So every player is investing or spending heavily on SGA. Once the market matures, say in a year or two, they should be able to cut off SGA significantly and it should come down to 20% to 25% of revenue.

Next R&D, which is about 50% of revenue, could scale down to 15% to 20%. So between SGA and R&D there are lot of levers for the company to pull and become profitable. But they will not be doing that for at least another 2 years. I expect in 2 years they could do $300 m in revenue and 20% operating margin. Depending on the dilution they could potentially make anywhere between 50 cents to $1. If they are profitable and $1 earnings and 10% revenue growth, is 15 to 20 x valuation too much? That is 50% to 100% from today’s share price. There are lot of risks here. But it is possible. Between their ability to grow revenue very fast and accumulated loss of $700 million, HDP has a floor underneath its share price.

And Bert responded as follows:

I would describe it as a distinction without a difference. By now, almost 80% of HDP’s revenues come from long term software-like agreements with customers. I would not think it likely that the market will mature quite as rapidly as the author expects, or that HDP gets to a 25% spend ratio on SG&A in the next 2-3 years. The market potential is far greater than that. R&D is not likely to scale down at the rates in the post any time in the near future. I would look for higher growth and higher spending than what the the author is forecasting.

Just thought it would be interesting for you.

Saul

4 Likes
19

I would not think it likely that the market will mature quite as rapidly as the author expects, or that HDP gets to a 25% spend ratio on SG&A in the next 2-3 years.

Agreed.

I was not clear when I talked about the two-year time frame. It was a thought experiment. The two-year growth is very visible, however, there is a much bigger runway of growth is available. The big data (Hadoop) addressable market is pegged at anywhere between 1T to 2T, of which in my opinion HDP can compete in about $10 to 20B. At the current run rate of losses if you project 5 years, while growth will be excellent, but we are projecting $1B losses. There is no cash flow in that growth to cover that. But the company is really in a very good long-term growth business.

HDP being a public company it has 4 choices in front of it.

One, Industry consolidation can help for sure, but there are only 3 viable players, how much more consolidation we are looking at? Two, continue to grow, incur losses and fund it through dilution, which I think is not viable given their current market cap, & valuation, note every company gets investor permission like Amazon. Third, taken out by a bigger company, who can leverage their existing sales infrastructure and reduce the costs significantly and make it profitable/ break-even immediately, while growing very fast. Lastly, HDP voluntarily slows down growth, where it can grow profitably.

The fourth model is what I described, it is a bit of a dream scenario. I do think HDP will slow growth enough to be able to fund the growth but not to the level that they want to get profitable right away. The more realistic scenario is someone acquires HDP and slashes its sales cost.

almost 80% of HDP’s revenues come from long term software-like agreements with customers
separately, this Q1, the company had 25% revenue through professional services vs 75% subscription revenue. Q4 it was 35%, and the last Q1 it was 33%. I still think Q1 doesn’t’ show the natural progression, rather a bit of anomaly. I haven’t looked into it, rather thought will visit that if the trend doesn’t reverse in the second quarter.

On the distinction without a difference, being 100% open-source has a value proposition but it is also very easy for the customer to move away. On the other hand, when you add 20% of proprietary software around 80% open source, you create stickiness. How much of value is debatable. But that is little more than “distinction without a difference”.

While SGA is going to be required for the growth, as their installed base grows, some amount of organic growth will start kicking in, i.e., existing accounts will grow, and this will require lot less SGA or in other words higher margin growth. I would model at least existing accounts growing 3 to 5% for next 5 years easy and depending on the accounts 10% growth is achievable. I think the real sweet spot is getting to $700 to $800M revenue rate where the company can get breakeven or some sort of adjusted profitability.

7 Likes
20

Hi Kingran,

Thank you for very valuable insights.

I was thinking a real example related to open source is RedHat (RHT)- which has been very successful the largest open source SW field so far - Linux - and growing consistently at ~15% / year over last 10 years… and its $2.4B revenue still gets PS of 6.7.

However, i do not have historical background on how this company evolved and what happened to other players in the space.

Would be very interesting to hear your perspective on how HDP or CLDR compare to RHT.

Also, FWIW - I just noticed that HDP has ~$23M of stock based compensation consistently for last four quarters… as a result they are much more closer to being operating and free cash flow break-even compared to what income statement would show.
Morningstar doesn’t show enough data on CLDR but seems like they are farther to reach cash flow break-even.

thanks
Nilvest
Long HDP and RHT
(small position each)

1 Like
21

I was thinking a real example related to open source is RedHat

Thanks for the kind words. I have never looked into this space.

I had an opportunity to have a close look at Electronic Arts, then a $1B company, adopting RedHat Linux for their online gaming in early 2000. And never bought any shares in RedHat, always worried about their valuation, while fully understanding their growth and recommending moving to Linux is the right long-term strategy.

In some ways, Cloudera is a bit closer to RedHat than HDP. Just from a growth point of view, I think you can use RedHat as the model.

While there are other Linux offerings, I have predominantly worked with RedHat and SuSe only. SuSe has gained a large acceptance in the NA market when it was acquired by Novell, it was far more popular in Europe due to its European ancestry.

Looking back to your past mistakes is scary, even more, difficult thing is to outgrow those fears. Living in Silicon Valley in 2000 you have seen many folks take a massive financial hit during 2000, is a serious mental scar and I have avoided growth investing (technology) completely. Perhaps a forum like this might have helped me to overcome a bit.