jad, excellent article. Thanks.
December 4, 1998
Is It Only the Beginning for EMC?
By Tiernan Ray
IN THE BEGINNING there was International Business Machines (IBM), and IBM owned everything. The technology of computing in the fifties, sixties and seventies was practically synonymous with Big Blue's products, in no small part because IBM invented the modern disk drive. Long before the storage business came to be equated with paper-thin margins and depressing cycles of boom and bust for companies like Seagate (SEG) and Quantum (QNTM), IBM transformed the industry with a product called RAMAC, a collection of vacuum tubes and aluminum disks floating on cushions of air. RAMAC for the first time let computer users store and retrieve pieces of data at random, the way a compact disk lets you play songs at will. It left the competition in the dust.
A generation later, IBM has an heir in another company with three initials, an extraordinary 19-year-old outfit called EMC Corp. (EMC). From its roots as a maker of replacement computer circuit boards, EMC has transformed itself over the past seven years into one of the fastest-growing technology companies that's never done business on the Internet. EMC is listed along with Seagate and Quantum in the storage device sector, but it is a whole different animal.
What EMC does is to take disk drives from Seagate and IBM, and package them together with some wild technology that makes fetching data off a disk drive super-efficient. The result is something called a "storage server." It's a separate computer, carefully engineered from the ground up just for data. These things can run two or three million dollars a pop. In other words, rather than being a storage company per se, EMC has become a computer systems company much as Cisco Systems (CSCO) is in routers and switches.
And like Cisco, EMC now has a host of companies riding its coattails developing technologies that may be critical to its future success. For investors, such small firms as Network Appliance (NTAP) and Computer Network Technology (CMNT) may offer the promise of heady growth. But it is hard to argue with the kind of success EMC itself has had of late.
On Wednesday, EMC's stock hit an all-time high of 80, giving it a total return of 192% in the year-to-date period. It's been a Dell (DELL) year for EMC, in other words. Forty dollars ago, you could have said EMC's stock is too rich, but today it's trading at a forward price-to-earnings multiple of 53 based on 1999's EPS estimate.
EMC closed up 4.5% on Friday as Morgan Stanley Dean Witter's Gillian Munson raised her 12-month target price to $100. Laura Conigliaro of Goldman Sachs wrote on the occasion of EMC's Oct. 20 earnings call that "EMC's Enterprise Storage Franchise Fortress Remains Unpenetrated." Wow! More specifically, Tom Lahive, senior analyst with Dataquest, says EMC owns 65% of the $5 billion enterprise storage market. Steve Milunovich with Merrill Lynch says he's confident EMC will do $10 billion in revenue in 2001, up from an expected $3.8 billion this fiscal year. (EMC is on a calendar fiscal year.)
In the quarter just ended, EMC sold a billion dollars worth of storage servers, up 40% from $700 million last year, and two-thirds as much as what Seagate sells in disk drives each quarter. EMC still supplies storage for mainframes, too. It's as if you took the storage part of IBM and turned it into a separate company. That's exactly how the market's valuing EMC these days. We wish we had bought in back in January, when it was trading around $35.
EMC has many unique qualities that remind us of Cisco. Sure, there's the team of aggressive young EMC salespeople who chase down prospects even on their honeymoons, as was charmingly related in a Wall Street Journal article earlier this year. And EMC develops and markets through a strong web of partnerships, including Microsoft (MSFT) and Hewlett-Packard (HP). That adds clout. Resellers and systems integrators help as well, such as, for example, Icon CMT (ICMT), the Internet service provider that caters to large business. Icon helps companies assemble EMC's storage servers into very large data centers using its facilities. Icon agreed to be purchased earlier this year by Qwest Communications (QWST), the aggressive nationwide communications company.
But the real key is software and services. By developing its own in-house software, EMC is gradually sucking into its storage servers other functions, such as network security. That gives EMC higher margins -- 50% gross, on average, compared to Seagate's 21%. And EMC has picked up on IBM's focus on services. The company's service revenue increased by 205% year-over-year in the latest quarter, partly through acquisitions, and the $56 million for the quarter represents 5.7% of revenues, double what it was last year. Having a service staff is what makes the storage server as reliable as mainframe storage has classically been. "EMC has won a huge part of its business based on a reputation for first-class service," says John Monroe, who studies the disk drive business for Dataquest. "Because of that, they really haven't had to reduce their prices that much."
And that service business is key, because EMC's most interesting battle lies ahead. Much of corporate America has been trying to kill IBM's mainframe for years, and the popular press has egged them on. The belief was that microcomputers, such as the humble desktop PC, and workstations from Sun Microsystems (SUNW), would eventually replace those musty mainframes. But IBM's machines have thrived, getting better and better and bringing in more revenue each year. The problem with trying to kill the mainframe was that the PC has a long way to go in acquiring the kind of reliability and raw strength of IBM's best machines. So like nuclear waste, all that data the banks put into mainframes years ago seem destined to stay in the ground. (These days the banks just pray their mainframe data doesn't go down the tubes when A.D. 2000 rolls around.)
What was needed was something to give PCs and other puny microcomputers a kind of impregnable vault, something as secure as IBM's mainframe disk drives. EMC can do that. Its computers, combined with off-the-shelf servers running Windows NT and Solaris, create a sort of forbidden city of data within companies, and one which is every bit as strong as the mainframe. EMC is what makes Bill Gates' aspirations seem even halfway plausible. As Milunovich puts it, "EMC's strengths help solve problems caused by Windows NT's popularity."
But having helped the PC guys battle IBM, EMC is now moving beyond them. It is building something new called the Storage Area Network, or SAN. Today most of EMC's computers merely connect to Compaq's (CPQ), Sun's or HP's computers. But the real future of storage, many say, is networking. Storage is no longer simply about saving files to a hard disk or floppy drive. With the rise of the Internet and email as a means of getting serious work done, analysts are talking about the increased need on the part of office workers to access large volumes of data on computers scattered across the Internet, maybe across an industrial campus, maybe on separate continents. The SAN would create a network between EMC's storage servers.
As the SAN emerges, EMC's software will probably suck more functionality out of HP's and Compaq's servers and into its proprietary computers. EMC's natural allies, then, are also its natural enemies in the long term. HP is a partner now -- it resells EMC's storage computers along with its servers -- but it could be a competitor in the future if it can get its act together. Sun already sees itself as a serious player in data storage. Compaq has products similar to EMC's. And Intel is working on new technology for server computers that will give Compaq even more firepower in accessing disk drives really fast. What's more, when Compaq picked up EMC's old nemesis Digital Equipment last year, it gained a first-class services business. Very smart. And then there's always Big Blue.
On EMC's side, there are a host of private companies trying to build networking products that work with EMC's storage servers. They include companies such as Brocade Communications and Gadzoox Networks, both headquartered in San Jose, Calif. These companies employ a technology called fibre channel, which is related to gigabit ethernet switches in development by Cisco Systems and others. Essentially, the same silicon chips feed the market for the gigabit ethernet and fibre channel, creating economies of scale -- meaning both networks split the cost of development.
What's interesting about these private companies is that because they know networks, they are in the forefront of building the SAN, which is, after all, just a special kind of data network reserved for storage, much the way local-area networks in buildings connect desktops. Now, because the SAN hasn't been built yet, Gadzoox and Brocade are running on empty, burning through venture capital money. EMC's working on the SAN day and night, but it could buy one of these companies at some point. So could HP or Compaq.
Aside from the startups, who's got the networking expertise to build the SAN? Our first choice would be Network Appliance. Like the name says, Network Appliance has for years been building a file server that plugs right into a network and makes it easy for end-users with different kinds of computers to share documents with each other. NetApp, as it's called, is technically in a different market but no matter: Storage networking is storage networking, period. Revenues are modest, an estimated $280 million this fiscal year, but they are increasing at about a 70% rate, and so it would appear Network Appliance's years of boring people with talk of file servers is finally catching on. At a recent price of 73, Network Appliance is close to its year high of 77 and trading at a forward P/E of 83.
Computer Network Technology, a tiny $169 million (market cap) maker of networking products located up in Minneapolis, Minn., is also helping EMC build the SAN. EMC partners with CMNT and promotes the company's products as part of the "total solution" for the SAN. Computer Network's main product line, dubbed the Ultranet Director, ships data between EMC's storage servers.
Why is that important? Imagine a situation in which an IT officer at headquarters needs to ship a copy of the master database to a remote branch office, say, to have a backup in the event of a massive system failure or an earthquake. The typical SAN configuration is not set up to transfer multigigabyte database files over so-called wide-area links, meaning network connections spanning hundreds of feet, even thousands of miles. But CMNT's products, which combine networking hardware with software, can transfer files long distance while making sure data is not corrupted along the way. Dataquest's Lahive thinks the company needs to integrate backup software from Veritas (VRTS) and Legato Systems (LGTO) into its wares to bring backup capabilities to the SAN. "They could create data backup that doesn't require any server computers at all!" says Lahive with enthusiasm. "That would be a breakthrough."
What will happen to EMC in 1999 as it tries to grow the SAN while companies worry about the Y2K bug? Given that Y2K is all about data on mainframes, EMC will be square in the middle of things all year. It could be that the cost of fixing the software problems will depress money spent on EMC's wares. It's more likely, though, that as we approach the millennium, many IT managers will realize there's no way to fix this stuff but to buy all new machines and just move the data out of the mainframe. EMC will be called in to bail out frustrated companies. In fact, '99 could be an interesting year in which the new client-server data centers of the future, and the SAN, meet their trial by fire.
|
