Enormous excitement over flash storage is driving a flurry of SSD acquisitions. Dot-com mania, anyone?
Those who cannot remember the past are condemned to repeat it, philosopher George Santayana wrote more than a hundred years ago. It's advice that data storage vendors would do well to take to heart.
Dot-com mania wasn't the only excess of the late 1990s; storage vendors snapping up SAN startups and rivals was another. Companies such as Veritas, Sun, IBM and HP, to name just a few, spent a fortune buying up small companies that had software, hardware or just ideas for SAN systems – and many times not real products.
SAN was just coming in to the market and was expected to dominate, and the big fish did not want to lose out to other players. These SAN startups were consumed into the larger companies with big design plans, yet only a few of the buyers did much with the companies they acquired, from what I saw.
Many firms that got bought ran into the big company NIH (Not Invented Here) group of engineers and got buried. So one side of the company was buying companies and the other side of the same company was killing the products. Even if those startups were not shunned by the big company’s engineering staff, I can think of very few products that were purchased by big companies that made it to market.
Lots and lots of money was spent during this period. I remember a company called TeraStor that burned through $850 million dollars (think of that in today’s dollars) in just a few years. Storage companies, file system companies, networking companies, disk drive vendors and lots of other SAN infrastructure or companies that supposedly could be infrastructure were purchased.
I remember doing some due diligence for some of the big companies that were buying these smaller companies. It did not matter what you said about the companies they wanted reviewed, they had money to burn and were going to buy companies no matter what you said. Fast forward 16 years to today, and lots of people seem to have forgotten the 1990s. Almost all of these purchases in the 1990s worked out very poorly for the big companies that did the purchasing, wasting money that could have helped during the recessions that followed.
Today, small companies in the storage industry, especially the SSD makers, are being purchased by big companies like piranhas in a feeding frenzy. We have seen Texas Memory purchased by IBM, Whiptail by Cisco, STEC and Virident by HGST/WD. And I believe this is not even close to the end of the buying spree by big companies. Additionally, we have seen investors pouring big money into companies like Pure Storage (a record $150 million).
So the question I want to ask is: does this buying fever make any sense? My answer: it makes no sense to me, and here is why. It might not be quite up there with tulip mania, but if history is any guide, much of this takeover money would have been better spent elsewhere.
The reason is that SSDs are a commodity technology. Making an SSD storage product is an integration project, not a rocket science project like designing a new chip. Some might disagree and I’m not saying it’s not difficult, given all of the hardware and software that must be integrated, but it is far more difficult to make an 18 nanometer NAND chip – including the design, lithography, and of course the plant – than it is to take a bunch of NAND chips and turn them into a storage product.
Of course there are issues with packaging, and yes there concerns about silent correction, error recovery wear leveling, and reliability. And there also concerns about the software and firmware, and hooking into a SAS or PCIe interface (folks, the SATA interface is not enterprise no matter what anyone tells you) and lots of other stuff. But this is integration, and far different than the complexity of making NAND chips.
It is also far less costly, as a NAND plant costs billions of dollars, whereas integration of NAND products or even integration of SSDs is far less costly. Whiptail, for example, last I checked was using SSDs from Intel, which are using a SATA interface (and I do not consider them enterprise). Yet Whiptail was bought by Cisco for $415 million, which is likely a great return on the investment made by the VC community, but is it for Cisco stock holders?
Is the software technology and storage packaging worth that kind of money? Could Cisco have done the same thing? Lots of questions, I would love to ask the people in Cisco that made this decision, and some of the same questions I asked over a decade ago.
But there were other SSD companies sold in the last 18 months and even more deals in the making according to market rumors. The SSD market consolidation is far than the 1980s when we had many disk drive companies, most of which were producing their own technologies.
The companies being bought are integrating technology from the big NAND players. The big NAND players (2Q13 numbers Samsung occupied 37.1% market share, Toshiba 28.9%, Micron 13.7%, SK Hynix 12.7% and Intel 8.6%), could decide to enter the market with products of their own design, and these vendors control the NAND distribution.
Some of these players already make their own SSDs. It would not be much of a stretch for them to compete in the storage market. But that costs money and most of these companies are not know as vendors in the enterprise market from a direct sales point of view.
Why do the companies want to sell?
Of course there is the obvious reason for investors to get their money. But small companies selling enterprise storage products have a high cost of sales, which requires a large infrastructure, which I do not think can be sustained in many cases.
When an integrated storage vendor walks in to sell products, they have a whole myriad of products to sell, from SSDs, to appliance storage, to block storage, to archival storage – you name it, they have it. A salesperson, a pre-sales person and a number of products to sell. This is far different than a single product vendor that might have one product or even maybe even a few solutions with different sizes and configurations, but not a set of products to address many different customer requirements.
Selling products is expensive and time consuming for most companies, especially small SSD storage vendors. You must have the right sales team that understands the customer requirements and sales process. If you are selling to small businesses you likely do not need this type of sales team, but if you are selling enterprise SSDs products to large organizations you need lots of infrastructure. You will need sales, pre-sales, marketing department, benchmark groups, and you will need to attend various industry conferences and shows and events. All this costs lots and lots of money and doing it with the wrong people will not improve your market or sales position.
The cost of sales and marketing for multiple products is a small incremental cost for additional products. So, for example, going to an industry show with one product is a high cost per product, while if you have 10 products you might need a bigger booth but your fixed costs will not be much higher. So these small SSD companies must sell like large companies, given the expectations in the enterprise.
I suspect many of these companies have a very high cost of sales for their SSD products. I would have loved to look at the balance sheet for what the cost of sales was and the margins that were required to keep the company in the black (or maybe they weren’t in the black). Of course with private companies, who knows if they are in the black or in the red? Only the acquiring companies will know that information.
So I have levied a great deal of criticism. What would I do if I were IBM, Cisco, WD or the next company making a purchase? First, I would look carefully within my company to see historically what has happened to other companies that have been purchased. Were the products successful, did we get a good return on investment? Did the new products come out on time? Was the vision we had when we made the purchase successful?
If you cannot answer yes to all of these questions for a high percentage of acquisitions, you need to ask yourself: what are you going to do differently? We all know that the definition of insanity is to do the same thing over and over and expect different results. So if you’re not making significant changes so that this time will be different, then you need to run, not walk, away from buying anyone.
I think we can all look at the purchases the companies listed have made and make our on value judgments. There are many alternatives to purchasing companies. From strategic marketing and sales relationships to OEMing the product, to developing a similar product approach with commodity SSD parts. Each of the companies buying enterprise SSD vendors has some significant engineering talent that goes far beyond integration of SSD products, so these companies could certainly develop their own products.
I really think that big companies are repeating the same mistakes that were made over a decade ago and making even more aggressive mistakes. A decade ago you had companies developing boards for new SAN switches, new RAID controllers with specially designed ASICs that were needed for parity generation and validation (all of this without PCIe so they had to develop their own buses), and lots of other technology that jumpstarted the SAN market space. Significant engineering talent worked on these technologies.
Yes things are different today, as much of the market is focused around commodity x86 technology, PCIe and software to build a product. At least for now there companies engineering PCI cards (networking, SAS, and SSDs), but today you can buy PCIe interface chipsets, 10 GbE chipsets, SAS chipsets and a portion of the design can be commoditized.
So back to my point, I seriously doubt that the STEC-Inc technology will be seen in HGST/WD SSDs, nor do I think that Virident PCIe cards will be commoditized by HGST/WD to compete with LSI and others. A Whiptail system will likely be put into a Cisco rack, but it’s not like Intel and Cisco are the best corporate partners, and we will likely see other SSDs put into the product. I really do not see Cisco as a company with deep storage talent and detailed understand of the storage market. Do they know how to sell to the storage team at their customers?
It all comes down to what I see as “the buying arms race.” Company X purchased some SSD company so company Y needs to do the same or they will not be considered a player. Are the idiots on Wall Street driving companies to make purchases to be considered innovative? I sure hope not, but I fear this is the driving force, sadly.