Dr. Rosenblum developed VMware’s virtualization software while working on a supercomputer research project with his graduate students at Stanford University, where he remains an active professor of computer science. In 1998 he went on leave from Stanford to launch VMware with four business partners, including his spouse, Diane Greene, who remains the company’s CEO. It wasn’t easy going. Back then, VCs had a hard time wrapping their minds around the business opportunity in Dr. Rosenblum’s software, which allows one server to do the work of many.
Today, VMware has 100,000 customers and is expected to sell nearly $2 billion worth of its products this year. And that’s after debuting on the New York Stock Exchange last summer in one of the most successful IPOs since Google.
Om spoke with Dr. Rosenblum in December about the history of the company and state of the virtualization market. Here Dr. Rosenblum talks about being a founder.
F|R: Some of our readers refer to founding as a “lifestyle,” not a job. What convinced you to table your very well-established career in science and academia to risk launching a startup?
Rosenblum: We were working on this supercomputer, trying to figure out how you could build scalable computers of a very big size. I wasn’t really that interested in high-performance computers — for one thing, you’d sell very few of them, other than to the government. What motivated me was wondering whether you could use these computers for something else. I had this idea about virtualization, that you could carve this one, big computer up and use it for a whole enterprise worth of computing. I was just more interested in working on something that would have a large impact on the masses.
One of the nice thing about Stanford is that the path is pretty well worn into entrepreneurship, so it didn’t seem like this radical thing to become a founder. If you had an interesting idea, there were plenty of people who you could talk to about it. Part of what convinced me were the grad students; they’d seen the Yahoo people go off and so they were excited about starting a company. I was sort of trailing along thinking, “Well, it sounds like a fun thing to try out.” But the deciding factor was when my wife, Diane, got interested in it. That made it incredibly easy for me.
F|R: Finding the right co-founder is a critical step. There is an unspoken rule in Silicon Valley that VCs won’t fund husband-and-wife teams. You did not raise VC money to launch VMWare, but what advantages did you and Diane Greene have as a founding team because of your status?
Rosenblum: Yes, we just did some self-funding in the beginning and then brought in some friends later — angel investors — and that was enough to do it. I can imagine that husband-wife teams can work out badly, but in our case, I had confidence in her dealing with the business side and she had confidence in what we were doing on the technical side and so we just partitioned up the company that way. And we didn’t really have any conflicts whatsoever. The nice thing about being married is that is gives you even more time to talk everything through, and it kind of consumed our life for a long time — maybe to the detriment of our children — but we just had great communication. It was a benefit, too, in co-founding with
faculty from Stanford academic colleagues (Scott Devine, Dr. Edward Wang and Edouard Bugnion). We knew were launching VMware with people we trusted.
F|R: What was the most difficult thing about VMware’s 10-year run to its IPO? And can you offer a piece of advice to founders just starting out?
Rosenblum: When we first started out the whole challenge was trying to convince people that virtualization was a good ting, what’s it’s good for — and trying to figure out for ourselves what it could be used for! There were challenges on the technology side and challenges about how to go to market. We took the approach that we wanted to partner extensively with [hardware vendors like IBM and Dell]. That ended up working out pretty well, though lots of people have made lots of money in shorter time that we did.
One thing I would say is important: Make sure there is more than one application for your technology. There were a dozen applications for VMware that we didn’t pursue. Some we did, and they didn’t work out. In the dot-com boom, we thought we’d sell our software to ASPs, who’d use it to manage other people’s applications more efficiently. Then all the ASPs went out of business. So we switched strategies to focus on selling VMware into enterprises, so companies could use it to run their own servers better.