The buzz around cloud computing has been a boon for the companies that deliver the enterprise services and run the server farms where the processing power and the data actually reside. But as the theme matures, the list of opportunities may be broader than you think.
To find them, it helps to cut through a little of the hype. Migrating to the “cloud” may be worth $35 billion to enterprise IT this year alone, but the move only passes the headache of having to host the software back up to the cloud vendor. Bigger clouds still need faster servers, more direct pipes to the Internet backbone and above all more storage space to hold all the data clients formerly kept on in-house drives.
In other words, mass-market computer manufacturers like Hewlett-Packard and Dell may be struggling as the customers who used to buy one or two servers at a time head for the cloud, but demand for true supercomputers is as strong as ever. And that’s where high-performance Flash drives come into play.
Data storage is already a $50 billion business as the wired world generates 2.5 quintillion bytes of information every day—enough to need 75,000 new terabyte drives a month just to keep up. And spinning disk drives are cheap, but at cloud scale they simply don’t stack well enough to stream the files as fast as modern applications need it.
When a small band of high-tech visionaries founded Fusion-io in late 2005, they knew that solid state drives (SSD) built out of Flash memory would one day be the best storage solution on the planet. And they were right. If you’ve played around with a MacBook Air or other SSD laptop lately, you know that these devices have come a long way since the thumb-sized 2-gigabyte “memory stick,” and they’re only getting better year after year.
The ability to read or write 500 megabytes of information—roughly a half hour of video—every second is probably close enough to instantaneous for any consumer computing application. But SSD is a technology that makes cloud operator’s scale possible. Multiply your needs by hundreds, thousands or millions of enterprise users working simultaneously and the traditional data center actually starts to melt down from its own mechanical heat. Every one added to a network cuts power use by 80%, requires practically no active air conditioning and supports thousands of times more user operations per second than HDD as well. Add more cloud customers, add more Flash.
This is true supercomputing territory, so it’s no wonder that Fusion-io’s current customer base is extremely concentrated: high-frequency trading firms where every millisecond counts, massive cloud operators like Facebook and Apple’s iTunes that need to handle millions of users all logged in at once to check their status or play their songs. The typical FIO customer buys a lot of drives—generally $10 million to $20 million worth per quarter—but you really have to be Facebook, Apple, Dell, Hewlett-Packard and IBM to move the revenue needle here.
Needless to say, a business like this is vulnerable if one or two accounts pulls back, and that’s exactly what’s savaged the FIO chart in the last nine months. Apple and Facebook have relaxed their drive orders to grow into the data centers they’ve just built. And while smaller cloud operators and the server makers themselves are stepping up, they have yet to fill the hole in the company’s growth curve or bottom line.
No profit in the recent quarter means no P/E, and declining revenue pushed FIO off the market’s growth screens. End result: a company that still has top-tier technology and blue-chip customers lining up to buy it, but shares are trading at only a fraction of their 52-week high of $32.63.
At this point a short-term relief rally—perhaps on a customer announcement or other news catalyst—seems overdue. If you believe FIO can be a game changer in the next generation of computing, watch for chances to build a position, but be careful going into the earnings announcement currently scheduled for August 5. After then, we’ll know a lot more about whether the trend has turned.
And if this particular piece of the cloud hardware story is too gloomy for your taste, tighter logic applies when looking at a name like LSI Corp. , which makes the processors that run Flash storage drives. Again, these are high-performance components normally associated with true supercomputers, but where FIO has found it tricky to move down market, LSI is already there with “hybrid” solutions that bridge the gap between new Flash and old hard drives.
Wall Street seems to think this is a rebuilding year for LSI as the company repositions its high-end products for another big sales push next year. But while consensus is for a $0.13 EPS when LSI reports on July 22, whispers and guidance indicate that the analysts may be aiming too low. LSI has delivered an upside surprise every quarter since mid-2011 and the spread has averaged a substantial 21%.
And with a growth-adjusted PE already well below 1, every cent the models are off now can snowball into some serious multiple expansion for the rest of the year. By that point, even the analysts agree that LSI will look cheap at today’s levels with a forward P/E under 10 and a growth curve on the way up. If it weren’t for the fact that LSI has already run up far ahead of the broad market since its last earnings release, it would probably be an easier call.
Either way, the cloud isn’t going to eliminate demand for high-end computing equipment. Mass-market servers and spinning drives are slowly going away, but the cloud vendors themselves will need enough Flash storage and SSD to replace every system their customers throw away. That’s the long-term story for companies like FIO and LSI—it’s just a matter of seeing how they get from here to there.