When Marc Andreessen and several colleagues from his days at Web pioneer Netscape Communications started a company called Loudcloud in 1999, they figured it was a sure thing. With Andreessen's cachet as the wunderkind who took the Web browser mainstream, Loudcloud, which aimed to provide computer-hosting services for companies racing to get a presence on the Web, looked like a can't-miss.
But the ensuing tech bust was unkind. Customers disappeared as their venture backing withered. And Andreessen, acting as chairman and chief technologist, and CEO Ben Horowitz were forced to sell off the hosting business to tech consulting giant Electronic Data Systems (EDS ). They were left with the software they had written to manage their data center -- and not a whole lot more.
"They couldn't get the economics of scale with what they were doing," says Donna Scott, a vice-president at market researcher Gartner. "It made more sense for EDS, someone with massive scale, to take over the hosting business."
A KEY PIECE.
But a funny thing happened on the way to near-ruin. Loudcloud became a little software company called Opsware (OPSW ) in 2002, just as tech gurus began talking about a new concept called "utility computing." The concept was simple, and Andreessen had been talking about it since the late 1990s: Turn a computer network into something that's as simple to use as an electricity or water utility.
It turns out that Opsware had one of the key pieces of software necessary to turn computer networks into something customers can pay for by the digital drink. Opsware was using the software to run its own data center. It allowed Opsware engineers to manage a room full of servers from a central point -- a key that supports the fledgling utility computing market. "Opsware/Loudcloud had the right idea underpinning for their business model all along,' says Andrew Schoepfer, president of Tier 1 Research, a tech research firm in Minneapolis.
Analysts call it real-time infrastructure-management software. What it does is fairly simple. It allows a company to manage all of its computer servers from one central control point. That means a tech manager can update software on various servers and allocate processing power as though hundreds or even thousands of computers are behaving like one giant machine.
It's not the only piece necessary for utility computing, but it's a big one. "If companies can't even standardize their server environments, how are they going to move to utility computing? They can't," says Scott. Tier 1 Research estimates the market for this kind of software is about $1.34 billion per year, but that estimate assumes massive discounts. The real size could be four times greater.
Moreover, this is just one sliver of the larger utility computing market that has drawn such diverse competitors as startup BladeLogic in Waltham, Mass., and just about every big hardware and software company, including Sun Microsystems (SUNW ), IBM (IBM ), EMC (EMC ), and Veritas Software (VRTS ).
On the surface, that would make Opsware's chances of survival iffy. But Andreessen & Co. are gambling they can grow to be a substantial independent software company by making products that work with the offerings of all the tech giants. It's a strategy that worked well for Oracle (ORCL ) in the database business in the 1980s when it decided to work with all comers and not tackle anyone head-on.
While Opsware is hardly the tech giant many envisioned Loudcloud would become, it's healthy. It has been cash-flow-positive for five quarters, revenues are expected to double to $35 million this year, it has more than 250 customers, and some analysts expect it to turn profitable within the next year. It has had two profitable quarters since it was founded. In its most recently reported quarter, which ended July 31, Opsware lost $813,000 on $8.6 million in sales.
And the 33-year-old Andreessen, as famous in the boom days for playing the role of enfant terrible as he was for being a programming whiz, has noticeably mellowed. "There are certainly good aspects to hypergrowth, like growing really fast," he quips. "But you make trade-offs. And every trade-off you've made comes back to haunt you when sales slow down."
Despite rumors that Opsware is a tasty acquisition target for Silicon Valley giants like business partner Hewlett-Packard (HPQ ) or Sun, Andreessen says he intends to stick around and turn this little company into a big one. He quickly adds that people who say they're looking to sell don't know what they're talking about. "We should immediately have them muzzled!" Andreessen jokes.
DONE IT ALL.
He points out that enduring software success stories, companies like Microsoft (MSFT ), Oracle, and Intuit (INTU ), were hardly overnight successes. "Bill Gates used to tell people he couldn't imagine Microsoft becoming a $10 million company," says Andreessen. "Then he said that about $100 million. Then he said that about a billion." In short, Andreessen is willing to wait this one out. With Opsware having more than $58 million in the bank and generating cash, he thinks it's financially stable enough for the future.
Some analysts also say Opsware isn't a likely takeover target simply because its market cap, hovering around $480 million, is way too high when you compare it to current revenues. Big companies will happily pay four or five times revenue for a company with good talent and good technology. But 10 times revenue isn't likely.
That suits Andreessen just fine. He has been on magazine covers. He got rich young off the crazy ride at Netscape, saw how a company can fall apart when it's growing superfast, and became more than a little bored during his brief stay at America Online (TWX ), which acquired Netscape in 1997. He's happy building a company from the ground-up.
But if utility computing should really take off and Opsware starts rocketing, he wouldn't mind that either.
By Jim Kerstetter in Silicon Valley