Interview With Mark Moshayedi, President and CEO, Stec

Mark Moshayedi served as a director of Stec since March 1992, as president since March 2007, and as CEO since September 2012. From January 1995 to September 2012, he was COO and CTO. From March 1992 to December 1994, he held various positions including president of R&D and SVP. He also served as the company’s secretary from January 1995 through March 2012. Moshayedi graduated with a BS in engineering from the University of California, Irvine, CA and an MBA from Pepperdine University. His wife, also from Iran, went to medical school in Spain, became a medical doctor there, then move to the U.S. but her family lives in Spain so the family come to Europe quite often. Mark has 15-year-old twins and a 4-1/2-year-old kid. He’s a music and volleyball fan, and collects classic cars, currently numbered at 12.

Are you CEO or interim CEO of Stec?
I am the interim CEO of the company.

When will you be replaced?
There is a case against my brother who use to be CEO and due to some regulations that exist in the U.S., he could no longer be the CEO. I’ve replaced him for the time being until that case is settled with the SEC.

Stec seems to be a family affair between three Moshayedi brothers.
What happened is that the company got started in 1990 by the three of us. One of my brother was mainly involved in the after-market side of the business. In 2007 we sold SimpleTech, which was the original company name, with all the products and everything, to a company called Fabrik, which then was sold to Hitachi, then to WD, who owns the SimpleTech brand name now. It was at that point that my brother retired. Manouch [Moshayedi], who use to be the CEO and is now the founder of the company, and myself have been around. There isn’t really other family member involved in the company. It’s due to the history that we’ve had with the company and the huge investment that we’ve made if we’re still there and really engaged in the operations. I wouldn’t say this is a family affair. I’ve had the most history with the company. From 2007 I was president of the company, and prior to that COO and CTO. We’ve been around and we’ve done everything from the technology side to finance, sales, marketing and everything else associated with the company.

Which country do you come from?
Originally from Iran.

There’s not a lot of Iran people in the storage industry. Am I mistaking?
Not many are part of the start-up landscape. But do you remember Lexar? It actually existed because of SimpleTech. We bought the business unit of Cirrus Logic, which was the solid state business unit, and we made Lexar. The people that we put there were all Iranians including Petro Estakhri and Mike Assar.

More than two years ago, Stec was the king of enterprise SSDs and it’s not the case anymore. What happened?
The competition. We started the enterprise SSD business in 2006. We engaged with EMC in 2007 and by January 2008 we made the announcement that flash drive was now qualified for the EMC system. That was the start of the whole enterprise SSD market. It wasn’t until October 2011, I believe, that competition was able to create a product that was competitive to our SSDs. As a result, there is a lot of competition, people are now engaged in selling SSDs to the major OEMs which used to be our major customers.

What failed with EMC?
Nothing really failed. It was more the desire to buy the cheapest product possible. At the end of the day it’s a matter of price. We’ve still got a lot of innovation in our product line. Companies that are willing to sell at no margin, or at negative margin, will have to make money. Our business tactics has always been to make money. We don’t want to ever sell something because somebody else is doing it. If they have more power, let them loose money. We are first and foremost in the business of making money. We’re not in the business of making SSDs, satisfying the storage market but to make money.

Who are now your biggest OEMs or customers and what do they represent in percentage of your revenues?
Over the last year we diversified our customer base and we’ve gone after enterprise customers directly. Last quarter over 15% of our sales came from direct enterprise customers. We have a few telecom customers in that area that are buying large volumes. On the OEM side we still do business with the EMCs, IBMs, HPs.

Your top three OEMs last quarter were EMC, IBM, Hitachi, accounting for 70% of your revenue.
That’s getting diversified though. But it’s decreasing. Our target by half of next year is to be 50/50. 50% from enterprise direct sales and 50% OEM. This percentage was over 89% during the same quarter last year. When we started, the only place where people could really use SSDs was the enterprise storage, not server side. It required a lot of hooks for the server side to pick up. And that may be one of the mistake that we’ve made. We focused so much on enterprise storage that we didn’t pay much attention to the enterprise server side. Fusion-io came in and they purely focused on the server side. These accounts require a lot of hand-holding. From what I understand today, Fusion-io has 180 sales people that just sell to enterprises. We have today less than 10. We’re adding a lot of sales and marketing people to be able to engage with those customers. We have the products, we need to go to those customers now.

Last quarter a non-OEM customer accounted for over 10% of our revenue. Which one?
It was part of the enterprise end-user accounts, it was a telecom customer in the U.S.

How many class action complaints your firm is facing now?
The main one is in the process of being settled. The terms have already been agreed to and it’s going to be settled. Some derivatives, parts of the larger one, are also going to be settled or go away by themselves because the main case is settled. We expect all the class actions to be settled by the end of first quarter. You cannot really have ten different cases on the same case, they consolidate each others and then you settle those. The lawsuit that we had is being settled and the company is taking a charge against that. The insurance company have also contributed.

I read in your SEC filing that IC devices, which are mainly flash chips, represent more than 80% of the component cost of your products. There are really only four flash chip manufacturers, Hynix, Samsung, Toshiba and Micron. Do you have any special contracts with these companies to be sure that you get your chips?
If you follow the semiconductor industry for memories, you’ll find that prior to 1995 these companies made money. Past that date all these companies have only made money now. Only Samsung consistently made money in this business. Micron, Hynix lost money. Toshiba sold all of their memories and thanks to their partnership with SanDisk they’ve made money. What happens in this industry is that you’re selling your fab space. You’re selling quantity, trying to sell the volume of the flash that you make. It’s a commodity product. The price may be 4 cents up 5 cents down but the product is available worldwide. We never had any availability problem.

One of your major shareholders, Balch Hills Partners, recently wrote a letter to Stec’s board of directors offering three alternatives for your company. What is your opinion on each of them?

1. A dramatic reduction in expenses, particularly in R&D
Over the last three years, Stec has been building it’s engineer organization. As you hire you get a lot of good people. Over the next few quarters we’re looking at reducing that. It’s a normal business practice, right-sizing your business for the amount of sales that you have. We were already doing what they suggested. As part of that, me and my brother reduced our salary to $1/year. This is the third time in over 22 years that we’ve done this.

2. A focus on the company’s core products (e.g. the SAS market)
We go where the customers tell us to go. When we started it was FC, everybody wanted FC. Then, one customer, Sun, wanted SATA. So we developed SATA for them. Then IBM came up and asked us to build a SAS product, we build a SAS product for them. Then Fusion-io started to make some noise and customers asked us a PCIe product with no host resources so everything could be done on the card. With Fusion-io there is practically zero management on the card, it’s all software. Our line of product grew from FC to SATA to PCIe to SAS, but yes, we do need to focus. Going forward we’re focusing mainly on PCIe and SAS. These are the two growing markets. Within PCIe there are three different factions:
– The custom, where we are today. No industry standards but fully SAS compliant. If you have a SAS software already running on your system you don’t have to do anything. It’s the premise of the SOP market, although we are not compliant to SOP standards.
– NVMe market which will be a peak later off in that market.
– And then with the SOP market we will be creating SOP parts.
The product line just within PCIe and SAS is significant. It’s said to go over 5 billion dollars over the next couple years. It’s a huge market and a huge opportunity for any company.

3. Explore all strategic alternatives, including a sale of the company
If there is a buyer there should be contacting us. You don’t put up a company for sale or you got no value. There has to be an interested party on the other side. If somebody contacts us, why not! If they have a buyer they should call us [laugh.]

When we take a look at the price of your shares today the company isn’t really that expensive.
No, it’s below assets value. We have a $187 millions of cash and over $200 million of other assets. We also have patents that are key in the business, which we believe many, if not all, enterprise SSD manufacturers infringe already.

How many people are there in the company?
Nine hundred people. Only me and my brother took down our annual salary to one dollar a year. And we’re putting some cost-cutting measures in for the company as well.

There has been some rumors of acquisition in the past: Dell, IBM, Seagate …
There has to be a reason why they would want to buy our company. Why would Dell be interested in acquiring an SSD company? Seagate haven’t contacted us. Western Digital always tries to buy things for nothing.

What’s your reaction if I tell you that Stec was too slow to transition to next generation products: cheaper MLC-based enterprise SSDs?
When we started shipping MLC-based SSDs to IBM, about two and a half years ago, the flash chips that we used was already in end of life. That generation couldn’t go to the smaller geometry flash. We had to wait for a new ASIC to come out, work with the new flash supplier, build technology around it and that took time. Developing technology offers surprises sometimes. It’s not always black and white. But today the products are all internally and externally qualified by customers. It’s stable. It got a variant of SAS, FC and PCIe off of the same platform. We can make drives up to 2TB using 24 nanometer NAND flash, cheap and stable. We’ve added our CellCare technology to it to extend the life of the flash. We use commercial MLC rated for 3,000 cycles, and by using our CellCare technology, we extend that to 40,000 cycles. It means that you can almost write unlimitedly on the SAS product for five years and you would still not wear it out.


But you also were late in SSD cache software.
We have a product called Enhance I/O, released already. The thought process behind this software was simple. IBM, EMC and our OEMs used our drives as a cache or as a tier in their storage subsystem. They quickly figured out that they were not going to be successful in selling their systems without a caching software. EMC announced FAST and FAST Cache, IBM came up with Easy Tier, Hitachi with their own equivalent. They only automated the way for the system to store its most used data. When we looked at the server market we saw this piece of software missing, a caching software that would run on the system automating this process, relieving the user from choosing what to put on the SSD and what to put on whatever is being attached to. We started a group of about a dozen engineers in the U.S. and 40-50 people in India to write this software. It runs on Linux, Windows, VM. We’re the only one who have passed the NetApp Testing Suite. We sell it separately or it could be combined with the installed SSD.

Are you going to continue developing your embedded market segment?
We’ve been in the embedded market since 1994 with CF cards, micro-SAS, 1.8-inch SAS. It worth about $15 million and we think next year it will do maybe $20 million per year. Yes we’re engaged in this market.

What is your roadmap?
Higher capacity, lower cost products. Small incremental engineering. We follow what the customers are asking for. No brand new development, just a repackaging of what is already available for that market. We also make a very unique product, ZeusRAM, it’s basically a DRAM powered backed up by super-caps. When the power goes away, it takes the data and put them into flash. When you boot-up, it goes from flash to DRAM. You can’t wear it out. It’s usually used for metadata. And we’ve been very successful with the Nexenta-based systems, used for the LogZilla product.

Are you going to enter into the subsystems market?
We are putting subsystems together as part of our sales strategy. But OEMs are building systems around SSDs and end-users are asking for system solutions. Subsystems solutions is a reference design and the customer can decide to buy it from us or they could go to subsystems vendors like Dell or HP.

Do you plan to have an all-SSD appliance?
Yes, but it will most likely not have our own software to start with. It will be equipped with a third-party software and the hardware will come from an existing supplier but it will have our SSDs and our configuration.

Will you sell a box with several SSDs, eventually flash arrays with compression, de-dup inside?
Those are the software that will be available from our business partners. We will be able to sell a Nexenta-based solution for example.

Because the all-SSD subsystems is going to be a huge market. Eighteen companies are building these products, generally start-ups and EMC also coming next year.
Big data is one of the reason for going in this direction, access to all the data all the time. The other reason is that de-dupe and compression work a lot better with flash than with HDDs, allowing inline compression and de-dupe. Also reducing your overall cost. With the advent of TLC for some of these markets with the right management, it’s easy to really reduce the costs below $1/GB. The key is building a systems that has some memory, ZeusRAM maybe for very quickly accessed data, MLC flash for data that are not written too often and some TLC for large amount of data, fast read but no writing. The majority of the system will be TLC based. Probably 99% of the world of big data is read but is written very little. Why burden the whole system with an SLC when you don’t need it? Our technology is entirely based upon getting the cheapest flash and the least amount of it. Most people probably don’t know that 80% of the cost of the drive is the flash. Certain companies that we compete with sometimes spend 120% of what we spend in the flash side of it. That’s because of the way they do the design and the over-provisioning to meet certain customer requirements. There are three parameters in the SSD to be concerned with, the over-provisioning has a direct correlation to the right applification and the amount of write/erase endurance overall on the drive. These three variables play a key role in how you construct an SSD and sell it to the market. We target no more than 20% over provisioning on a drive when we sell it. Certain competitors have 70% over-provisioning. Using CellCare provide more endurance out of a flash and we create a lot of parallels inside our SSDs to reduce over provisioning. I believe that we have the most cost-effective enterprises SSDs that’s being manufactured.

But your prices are higher.
Yes because we compete against company that are willing to sell without making money. Hitachi for example make money on HDDs. We are on the business of just selling SSDs.