STEC Revisited: Will It Go from Good to Great?

I got this wrong. However, now that the emotions are out and the stock has settled into a new trading range, I thought I would take the time to analyze what happened and figure out where to go from here.

What happened at the Q3 Earnings announcement?

While STEC reported numbers in-line with consensus expectations, the only positive standout was Mach8 revenues which grew over 50% sequentially. The EMC revenue being sell-in (as opposed to sell-through) and contractual was not meaningful to analyze, but non-EMC Zeus IOPS was a disaster, which meant [[IBM]], SUN (JAVA), etc all underperformed. A miss on total revenues was averted through an un-anticipated jump in the “Other” revenue line with generous help as I understand from flushing the Dell (DELL) product.

The Company also guided down Q4 revenues to a range of $101 to $103 Million. While this was not bad enough, STEC then announced that the sell-through of the EMC contract revenues of $120 Million which was to be exhausted by end of Q4 is going slower than expected and that EMC might hold inventory into 2010. By not being able to provide any magnitude of the inventory level or actual sell-through progress, they created complete uncertainty on the run-rate revenues, growth rate and margins.

The stock market reaction was decidedly vicious, but more so due to two major reasons:

1. Uncertainty over underlying demand and growth rate: While there was some chatter that this was all due to competitive pressures, even a cursory analysis shows that it was not the case. The slowdown was all in the Company’s Zeus IOPS product (goes primarily into storage arrays) which currently has minimal competition. On the other hand, the Mach8 product that goes primarily into servers where there is lot of competition from FusionIO, Intel, etc actually grew over 50% sequentially, albeit from a small base. A sequential drop off in SUN revenues was anticipated, but is a one-off issue. So what happened with EMC, IBM, HP (HPQ), HDS, etc? Was STEC and enterprise SSDs the enterprise computing equivalent of Crocs (CROX) – a flash in the pan with no secular growth drivers? The answer is an emphatic NO. The need for Tier-0, a persistent storage tier that exists between fast, but expensive and volatile (non-persistent) DRAM cache and inexpensive and persistent, but very slow HDD is very much needed in the enterprise. Once NAND prices fell below DRAM few years back and the spread started widening, the economics was created and the nascent technology got converted into a robust, working product with compelling benefits for the end user and good economics for the vendors. The enterprise OEMs who adopted STEC and the startups and large companies that are chasing STEC are not all wrong. All what happened was the technology and expectations just got ahead of the supporting infrastructure catching up. Here are some of the reasons:
   1. Lack of management software: The initial market ramp for enterprise SSD’s came from early adopters, primarily high performance computing customers who live and die by speeds and feeds (the same guys who initially drove every cutting edge technology like FC, IB, 10G Ethernet, Blade servers, etc), who are happy to experiment on their own and make bleeding edge technology work in their data centers. STEC and their OEM partners also probably benefited from a lot of customers buying pilot installations. However, the hockey stick ramp comes from main stream adoption where system administrators want management software such as automatic data tiering that will automatically move hot and cold data between the various tiers of storage. Unfortunately, among STEC’s partners, only Compellent (CML), which is too small to make a difference for STEC has such software in their arrays. While IBM has a semi-automatic migrator in the SVC product just released in November, the fully automated “Automatic Data Relocation” will not be released till 2010. EMC’s initial version of automated tiering called FAST v1.0 is expected to be released any day now (probably this month), while the more granular, block-level version (v2.0) is not expected till later in 2010.
   2. Heavily over priced: At the same time these OEMs are trying to create this market, as only vendors like EMC and IBM can do, they heavily over priced enterprise SSDs in the market place. In their defense, it is hard and takes time to figure out the price elasticity of a brand new product with such incredible performance improvement and value proposition and so it is always best to price high and then come down to figure out the optimal market clearing price. The list price of an IBM 69GB STEC based SSD is $13,235. At a typical 35% discount, IBM is selling to the end user for around $8,600. The IBM 69GB SSD is actually a STEC 128GB SSD that IBM formats down for wear leveling. Using the spot $4 price for the SLC NAND and my estimates for the cost of the Xilinx FPGA, other components and manufacturing and other fixed cost allocation, I estimate that STEC’s cost for the 128GB drive is around $1,000 and I estimate that in Q3 2009 they sold it to the OEMs at a 60% gross margin or around $2,500. This means IBM is making a whopping 71% gross margin at an end user price of $8,600. While customers have always been willing to pay a premium for performance (a SAS drive can be 2 to 3x the price of a SATA drive), the current premium for enterprise SSD’s is probably way too high to bear except perhaps for some of the high performance applications on Wall Street for instance. Since EMC takes so much of their cues from the financial services industry (their largest vertical), it is possible that they extrapolated demand based on the successes and price elasticity in that vertical which probably did not pan out as expected.

Finally, in the case of EMC, since they experienced a 160%+ growth in SSDs in Q2 over Q1, they were probably concerned about going into allocation with STEC in the second half (especially with IBM rolling out STEC based SSDs) and decided to go with a large $120 Million order.

2. Concern over management integrity and credibility: Whether management knew about the demand/inventory issue in advance or not, the market found it too coincidental that top management made such a substantial sale of stock in the very quarter they blew up. At the same time the potential question of integrity was being debated, they made it worse in the earnings conference call by not appearing to have concrete answers about the inventory at their largest customer who will generate over 60% of the revenues in the second half.

Where do we go from here?

From a fundamental standpoint, all of the reasons I have articulated previously are still intact:

1. The market is much bigger than anyone is projecting. There has always been a big gap between the expensive, high performance, but non-persistent DRAM cache and the inexpensive, highly persistent, but low performance spinning hard drive. SSD’s fill this gap and over time every enterprise storage array and high end servers will have some SSD storage standard. This is a secular shift and we are in the early stages.
2. The enterprise SSD is not a commodity and STEC is the technology leader. The NAND chips that go into an enterprise SSD is a commodity and the hard drive that the enterprise SSD is replacing is a commodity, but the enterprise SSD is not. The intellectual property in an enterprise SSD is in the controller and a lot of sophisticated firmware goes into it to realize the right balance of performance and longevity, while ensuring the utmost level of reliability, availability and data integrity. STEC has been it at it longer than anyone else and those years of production burn time in the field keeps them well ahead of the competition. In my experience with similar products requiring many 9’s of reliability, anyone can create a product with 99.9% reliability within 12 months. However, to add each extra “9” of reliability can take as much as 6 to 9 months or more and requires a lot of field time to get there. Witness the number of high severity firmware bugs Intel SSD’s have had just in the last 12 months.
3. The competitive concerns are over blown. While it is certain that every OEM will have a second source, it can take a long time to qualify, test and benchmark a new product at a tier-one OEM. Enterprise OEMs like EMC and IBM know my previous point that just because a startup like Pliant or a major vendor like Seagate has released a product, it still takes a long time to get the five 9’s or more of reliability. Finally even when the second sourcing happens, the impact of the second source will be minimal to STEC over the long term.

Unfortunately, these points were valid even prior to the blow up last week. The stock is weighed down by uncertainty over the revenue run rate, growth rate and margins. Valuations are meaningless (7x consensus 2010 EPS estimate of $1.90) when investors do not believe the numbers. Technically the stock is broken and unfortunately fundamental investors will not support it till the uncertainty starts to improve.

Q4 is more or less meaningless at this point (because EMC is sell-in and already baked in at around $66 Million gross) and the focus will be on the direction of Q1 and full year 2010 estimates. Currently, the range for Q1 is $0.29 to $0.45 with an average of $0.38. The range for CY 2010 is $1.50 to $2.45 with an average of $1.90. The question is over the next few months whether Q1 consensus will start drifting closer towards $0.29 or move north towards $0.45.

While it is still too early to call, my feeling is that it is less likely to move towards $0.29 (which implies around $80 Million in revenues, with EMC likely around $40 Million or less and an overall 48% gross margin). Note that the consensus for Q1 is still not yet firm because not all 12 or 13 analysts have logged in their Q1 estimates yet. I believe that Q1 has a better than even chance of exceeding the consensus. Most importantly, EMC will have to fix their inventory issue and get a grip on steady state demand and growth rate. Here’s why I believe we will get there in the next few weeks:

1. EMC will flush the excess inventory: EMC is notorious about using all means to work off excess inventory. This time is no different. But more importantly, this is not a matter of just excess inventory for EMC. EMC has been behind solid state for a long time (the last time was in the late 90’s but the economics were not on their side then as it is now) and they will put all their might behind this to make it succeed this time. An EMC sales rep can sell dozens of products in a quarter to meet his quota and earn his commission. Thus it is amazing what a little bit of management focus (make it mandatory to quote SSDs on EVERY Symmetrix or high-end Clariion sales opportunity), management monitoring, price cuts and targeted sales “SPIFFs” can do to move a particular product. EMC is doing all of the above for enterprise SSDs this quarter. As illustrated earlier, there is such a price/margin cushion that EMC can cut prices by 20 to 30% or more and still make very good margins (they will most likely start with smaller price cuts). On the sales SPIFFs, STEC is generously helping out. STEC announced that their MDF for Zeus IOPS in Q4 will be $2.4 Million. I have no idea how this will be broken down. For the sake of argument, assume that around $900K will be spent on marketing programs (brochures, seminars, webinars, lunch & learns, etc) and the remaining $1.5 Million all for sales for SPIFFs. Assuming STEC does $75 Million in Zeus IOPS sales in Q4 and the OEMs sell it at a 50% gross margin – that is, $150 Million to the end users – the $1.5 Million translates to 1%. For an enterprise hardware sales rep making say, 3 to 4% commission at quota, an extra 1% is at least 25% or more in take home pay for the quarter. Sales reps are coin operated and they will figure out every way to move this product. Based on this I believe that a substantial portion of the EMC inventory will be worked off in Q4 itself, and I estimate that normalized run rate revenue from EMC in Q1 will be around $50 Million.
2. Other OEMs will improve sequentially from the Q3 low: While the SUN situation will continue to hurt for the foreseeable future, I think IBM can show good sequential improvement with the new DS8700 storage array released in late October, the new SVC in early November, and the new p595 in late Q3 all with STEC drive support. Note that the STEC MDF is available to all OEMs and it is very reasonable to assume that MDF will continue into Q1. Other new OEMs like LSI are just starting to kick in.

While it is difficult to call the short term with all the uncertainty over this, and I concede that there is nothing preventing it from dropping few more points on a technical basis, I believe that this stock can be a good investment with a 6 to 12 month time horizon.

I believe that STEC has earnings power of around $2.25 in 2010 and $3 in 2011. That is a 36%+ growth rate over the next two years. Unfortunately, STEC has never enjoyed a P/E multiple befitting their perceived growth rate in the past. If they can start getting a valuation multiple that reflects their growth rate, this stock can go from being a good investment to a “great” investment. However, that will require some major changes at the Company level. For what it is worth, here are some suggestions for management and the Board:

Organizational improvements:

1. Improve the bench strength at the executive level: The Company is growing way too fast to be operating with such a small executive staff of a CEO, a CTO doubling as a COO and a CFO. The Company needs to bring in at a minimum a world-class EVP of Marketing, an EVP of Field Operations, and a General Counsel.
2. Improve marketing: This will come from having a good Marketing executive. STEC has less brand recognition than even FusionIO, a much smaller company. If STEC had been doing industry and end user marketing helping end users understand the benefits of using SSD’s in the enterprise and the differentiation of their technology and approach, sales could be much better. This will greatly help the OEMs sales organizations and they expect it from vendors such as STEC.
3. Improve field operations: While STEC has done a remarkable job in signing up almost every major systems OEM to sell their products, in some respects it is more difficult to retain the OEMs and sustain and grow the revenues year after year in every channel. Competition is coming on strong and they will use every means to knock the incumbent out. This will require getting much closer to the OEM, meeting their every need, exceeding all expectations, helping in marketing, etc which is all about sustaining relationships. While technology and product got you in the door, a dedication to sustaining and constantly improving customer relationships is what will grow the OEM revenues. In fact, if STEC had that, we probably would not have had the “surprises” last week.

Improve Board governance:

1. Make the Board more independent: Of the 7 member Board, only 3 members, namely Mr. Bahri, Mr. Ball and Mr. Witte are really independent members. Mr. Moses was the previous CFO of STEC and is not independent as acknowledged by the Company. Mr. Colpitts is considered independent by the Company according to its 2009 Proxy Statement as follows: “In determining that each of the non-employee directors is independent, our Board of Directors considered, among other things, the types and amounts of the commercial dealings, if any, between us and the companies and organizations with which the non-employee directors are affiliated. In making this determination, our Board of Directors considered the fact that Mr. Colpitts is the Managing Director and Global Head of Technology Investment Banking at Deutsche Bank. STEC engages a different division of Deutsche Bank to provide cash management services. The fees paid to Deutsche Bank for cash management services do not benefit the investment banking division or Mr. Colpitts. Our Board of Directors determined that this relationship would not interfere with the exercise of independent judgment by Mr. Colpitts in carrying out his responsibilities as a director.” However, Deutsche Bank’s Investment Banking unit was the lead underwriter for the recently completed secondary offering and unless Mr. Colpitts has changed jobs, he does not now qualify as an independent Board member by STEC’s own definition. The Board should be reconstituted to be comprised of a majority of Independent Board members.
2. Add Board members with more industry and executive background: There are currently no non-employee Board members with background in operating enterprise IT companies and no Board member with experience in running publicly traded companies as a CEO. It would be a positive to have Board members with relevant industry experience and CEO experience who can coach and mentor Mr. Manouch Moshayedi.
3. Separate the role of CEO and Chairman: Much has been written on this subject over the last few years, but especially in light of the concerns the investment community has about the integrity and credibility of the management team, it will be a good idea to bring in an independent director as the Chairman of the Board.
4. Eliminate poison-pill provisions in the Company charter: Below is an excerpt from the Company’s latest 10Q that the Board could address in fixing:

We have taken a number of actions that could have the effect of discouraging a takeover attempt. For example, provisions of our articles of incorporation and bylaws could make it more difficult for a third party to acquire us, even if doing so would be beneficial to our shareholders. These provisions also could limit the price that certain investors might be willing to pay in the future for shares of our common stock.

These provisions include:

limitations on who may call special meetings of shareholders;

advance notice requirements for nominations for election to the board of directors or for proposing matters that can be

acted upon by shareholders at shareholder meetings;

elimination of cumulative voting in the election of directors;

the right of a majority of directors in office to fill vacancies on the board of directors; and

the ability of our board of directors to issue, without shareholder approval, “blank check” preferred stock to increase the

number of outstanding shares and thwart a takeover attempt.

In addition, provisions of our 2000 Stock Incentive Plan allow for the automatic vesting of all outstanding equity awards granted under the 2000 Stock Incentive Plan upon a change in control under certain circumstances. Such provisions may also have the effect of discouraging a third party from acquiring us, even if doing so would be beneficial to our shareholders.

Finally, from a tactical standpoint it will be a good idea to reinstate the stock buyback. There is still $5 Million of authorization remaining from the previous repurchase plan. The Company has $133 Million of cash on the balance sheet and is on track to generate another $30Million+ of cash flow in Q4. It will be a good idea to expand the stock repurchase plan by at least an additional $25 to $30 Million. The number could be higher but I have no idea how much of the cash is in the United States versus in other countries.

Price	Shs Out	Mkt Cap	Float	LT growth	Cash/shr	Oper Mgn
$13.50	52.2MM	$705MM	63%	30%	$2.55	31%
2008A eps	2009E eps	2010E eps	P/E 09E	P/E 10E	Revs 10E	EV/10Rev
$0.31	$1.65	$2.25	8.2x	6.0x	$480MM	1.2x

Disclosure: Long STEC

Comments

[User 499802:]

GARP: You really do some great analysis, but of course you lost a lot of credibility with your strong buy going into the last quarter. Have you ever thought that maybe STEC should be looking at diversifying their product line and customer base in order to put an independent floor under their overhead. SNDK was near bankrupt when Eli figured out that his technology was too far ahead of the market, so they went into other areas and waited until the market caught up with them.

Nov 16 10:32 AM

[rd4sndk:]

499802... I disagree with your essessment of Sndk. Sandisk financial problems weren't a result of its technology being ahead of the market. Sandisk failure stems from their inability to bring to market the right technology and the right product--namely SSD, OTP and 3D. Sandisk spend more time developing CE products (mp3 particularly) and aquiring msystems. All of these efforts did little to differenciate Sandisk from commodity markets while providing time for its competitors to catch up (albeit with lower profit margins) but never the less comparible or better pricing. Sandisk inability to stay ahead of the competition was its own undoing. STec should heed the message that your competition is more formidable than you are lead to believe by your initial success.

Nov 16 01:07 PM

[kingfrogcash:]

Great article. Thanks.
I thought the Cable TV vendors for VOD had a the software for their long tail. Newly released movies in RAM to old movies on disk.
Having OEM's qualify STEC isn't enough. Ordering the product cures all. Flash needs to be JBOD. Commonplace not specialized.

Nov 17 06:57 AM

[user1010101:]

Sorry GARP - but you have last too much credibility. Anything you say now just sounds like made up stories (i.e. your belief that EMC is pushing SSD's by offering incentives). I think you need to step back and recognized everything you initially thought was incorrect - it is clear SSD adoption is not taking place as rapidly as you thought. HDD storage is cheap and short stroking drives to increase performance is understood by IT departments. It will take time before comapnies decide to switch over to a new device - prices in $/Gb will need to come down in order to convince people.

I am glad you have suggestions for the company but you should have presented all of these risks in your initial reports. Clearly, there were conflicts of interest on the board and management was weak.

You became too attached to STEC and didn't think objectively.

Nov 17 09:03 AM

[sliman:]

Lawsuits will dog this company for the next year. No hurry to buy.

Nov 17 10:11 AM

[rkmsys:]

Great analysis! but a questions I'll be happy to see answered:
according to my past exposure to STEC business, they had significant part of their revenues in DRAM modules. There is completely no reference to that part of business in analyzing their (current, and) future results.
Is this gone? And if not - what impact this should have on combined results?

rkmsys

Nov 17 01:03 PM

[Scax:]

Thanks for the fact-filled analysis update! -- and noone is going to call them right all the time, so no problem on the past as I just bought in at 13ish with some downside risk, but lots of upside potential. The lawsuits have already been mostly discounted by the market, but may play a future role in valuations, not in the near term. They take time to mature and if this stock does rise as predicted, they become a non-issue for the most part. I really believe in this technology and agree with your recommendations to the company leadership. They are kind of like a brilliant teenage clique who are in the doghouse for reasons they'll figure out soon enough and when they come out with just a few neat improvements, the stock can't but do great things, in reply.

Nov 17 04:47 PM

[merigoldman:]

Has anyone heard anything coming out of the analyst day yesterday?

Nov 18 10:12 AM