In search of the next big thing
- By Susan M. Menke, William Jackson
- Jul 30, 2003
Marc Andreessen: Inventor and chairman
Marc Andreessen and friends invented the first Web browser, Mosaic, while they were undergraduates earning $6.85 an hour at the University of Illinois at Champaign.
But Andreessen is not an enthusiastic alumnus. He confesses to a lingering bitterness about the university's refusal to grant him rights to his phenomenal brainchild.
After graduating in 1994 with a computer science degree, Andreessen and some colleagues headed west to Silicon Valley, where he and SGI founder Jim Clark put together Netscape Communications Corp. They reworked the Mosaic concept into Netscape Navigator. The initial public stock offering made Andreessen an instant multimillionaire.
His subsequent job as chief technology officer of America Online Inc. ended less than a year after AOL's $10 billion acquisition of Netscape.
Andreessen moved on to co-found Loudcloud Inc. in 1999. Last year he rebranded it as Opsware Inc. of Sunnyvale, Calif.
The company claims that Opsware System 3, with service automation modules and a knowledge base, can deliver server-to-administrator ratios as high as 100-to-1 by automating the time-consuming tasks of server management. That, according to company officials, eliminates the unintentional maintenance changes that cause up to half of all system outages.
GCN chief technology editor Susan M. Menke and senior editor William Jackson interviewed Andreessen during a recent trip to Washington. GCN: You've said you haven't written any code in 10 years. Why?
ANDREESSEN: I ran out of time after we started Netscape. In a corporate environment you have to be able to work as part of a team, and the team expects you to be there. It stopped my carpal tunnel syndrome. Now I have carpal thumbs from my BlackBerry. It's attached to my thumbs.GCN: What was it like when you were inventing Mosaic at the University of Illinois and the National Center for Supercomputing Applications?
ANDREESSEN: We were all students, working under National Science Foundation funding. It was a little bit ironic, but that money was intended to centralize supercomputing for scientists under the theories of the mid-1980s.
By the time I got there in 1989, fast computers were expensive. They had $20 million Crays. They put them in a central location, like Illinois, where nobody would want to go, but they wanted scientists in other locations to access them. At the time that was a radical idea. So they started us off to write the software that the scientists would use to reach the supercomputers.
That has led directly to movie animation in films like 'Jurassic Park' and advances in scientific visualization. The computer animators themselves came out of that.GCN: What was your core group?
ANDREESSEN: At first my team was working on how to actually use the supercomputers. Eric Bina and I worked on something called Collage for the X Window System and Mac OS'an interactive way for scientists to collaborate on data visualization. This was all funded by NSF, over the Internet as it was then.
That eventually led to a small group of us coming up with Mosaic for X, largely without much management knowledge or sponsorship. Last April was the 10th anniversary of Mosaic.
At that time there were hundreds of thousands of users'defense contractors and the government. The agencies had IP networks, but typically large companies had only LANs. They didn't have IP or e-mail'it was largely confined to academics and researchers.
We released Mosaic to a small number of people. Then we began to see a phenomenon, a viral kind of growth, where the first 10 people would each pass it along to another 10 people each, and so on. We were able to track it in downloads, which grew to thousands and then hundreds of thousands, and eventually by the requests for technical support.
We were a bunch of undergrads and had no clue how to handle the technical-support requests, which were all coming to me.
When we graduated and left the University of Illinois, we went to Silicon Valley. It developed the way it did because of Fred Terman, who I call Hewlett-Packard Co.'s professor of technology. He had a well-developed theory of entrepreneurial companies being founded by universities. Silicon Valley is what it is because of Stanford University. When a company like a Cisco Systems Inc. or a Silicon Graphics Inc. or a Sun Microsystems Inc. takes off, the university benefits.
Illinois, however, was not one of those universities. They were very control-oriented.
We said, forget it. These guys are going to be sorry.
From the university's viewpoint, they didn't have this entrepreneurial tradition. They viewed anybody leaving as a threat to their ability to continue getting research funding. If we weren't there to keep doing the programming, they wouldn't continue to be able to get the money.
They were already so mad that we were leaving, and they were so difficult to work with, that we didn't ask for rights to our software. We left our code there. We said, we'll start over from scratch. We built Netscape Navigator.
In our early days, a company called Spyglass Inc. licensed Mosaic from the university and turned it into a commercial product called Spyglass Mosaic. So in effect we were competing against our own software.
Spyglass eventually made a deal with Microsoft Corp., which then held a press conference and announced it was going to give away what has become Internet Explorer. That drove an immediate stake through the head of Spyglass. The university began telling our customers it was going to sue us over the rights to the intellectual property, but eventually we came to an agreement.
Today, if you look under About Internet Explorer, you'll see a credit to NCSA Mosaic. You'll also see, 'Distributed under a licensing agreement with Spyglass Inc.' So it's our software underneath.GCN: And the rest is history. How did your current creation, Opsware, evolve from Loudcloud?
ANDREESSEN: Loudcloud started out outsourcing, doing managed Web hosting, running the infrastructure behind dot-coms. We developed the software to do that job.
Then a number of government agencies came to us and said they could not outsource their operations, but they wanted to buy our software. At first we said no, but we wound up selling our own automation software, which we ourselves had written to run our business. We ate our own dog food for several years. EDS Corp. acquired our hosting business for $64 million.
There's been huge, huge growth in computing. Organizations have to hire one person to take care of a certain number of servers, or one database administrator to handle so many databases. Opsware automates that so that fewer people can handle more servers or databases.
Google.com has about 20,000 servers all running the same code, with a massively distributed search engine. It's a lot like the rendering farms at Pixar Animation Studios. They use grids of 10,000 or 15,000 computers at once for movie animation in 'Finding Nemo' and the 'Toy Story' movies.
The typical government agency, on the other hand, runs large numbers of servers, all with different sets of software. Some have Microsoft Windows, some have Oracle.GCN: What's the next big thing?
ANDREESSEN: That's the trillion-dollar question. It took the world about 25 years to make the transition from the mainframe environment to client-server.
It will take another 25 years to make the transition to browser computing, so there is another 15 years to go in that process. What's coming after that, we'll see arriving somewhere in the next few years.GCN: So in spite of everything the government is doing to Web-enable systems for
e-government, there are 15 years to go for the browser?
ANDREESSEN: The browser has made it so easy to access applications over the network. The server does the processing, the browser does the display. I don't know what the next transition will be. Maybe to smaller PC devices, like phone-personal digital assistant combinations.
Or maybe to networked radio-frequency ID tags. Large companies and the military are using RFID tags on every item for logistics. We've seen that succeed in the war in Iraq.
Imagine that every box of detergent in the supermarket has its own chip that can not only say where it is on the shelf, but what's inside and where it came from and when it was manufactured, at a penny per tag.