Salil Deshpande is a managing director for Bain Capital Ventures who specializes in open source start-ups. He recently left Bay Partners after seven years and investments in a number of open source companies such as SpringSource (Java application servers), MuleSoft (enterprise and cloud integration), Engine Yard (Ruby on Rails), G2One (Groovy and Grails), Sonatype (open source component management), Dynatrace (application performance management), and ZeroTurnaround (making Java faster), among others.
Deshpande talked with ITworld about the business model for open source companies, what he looks for in a start-up as an investor, and how open source software infiltrates the enterprise. You started at Bain Capital three months ago, but you still have connections to your Bay Partners investments. It's like you're working two jobs. I am working two jobs. I'm still on the boards of some Bay companies, but I'm going to transition off of them. I did a lot of investment at Bay Partners. Over seven years, I invested $85 million in 20 companies. The bulk of those were done during a four-year period. Most of them are open source. How did you become the open source VC guy? I was a practitioner in the field since 1991. I got an undergrad degree at Cornell, got an MS at Stanford, and went to work for Sun Microsystems. I wrote code that went into Solaris, which was not open source, but it was Unix -- it was the closed-source competitor of Linux.
So you were a programmer. I was a programmer. And a lot of the stuff that becomes open source often is not applications, it's the infrastructure software. Things like databases, data grids, caching, performance monitoring, all these types of things we lump into this category called software infrastructure. It's not application software, it's infrastructure software -- software that makes other software work. Tell us about your "floorboard" theory of enterprise software. Look at what happens with SaaS. Let's say there's a department with five salespeople and they want a CRM system. This is obvious now, but 15 years ago it wasn't obvious. They say, "Let's not install any software, let's not hear a sales pitch from a guy who comes in wearing a suit about how great their CRM is. Let's just pay the $50 a user for Salesforce and just get five users up and running." And then five users start using it and they get value. They hire more people and pretty soon 15 people are using it. Pretty soon the department next door is saying, "Oh, let's do what those guys are doing." And pretty soon it's 200 people, and if this is part of a company, there can be a high-level conversation between Salesforce and that company where they say, "Look, you've got a couple thousand people using our stuff. Do you want to do an application deal?" And that's what I mean by application software coming through the floorboard. The same kind of thing happens in open source with infrastructure software because you're getting the same kind of dynamic. Your customer for infrastructure software is the engineer, or the developer, or the guy in your IT group or engineering group who is building software, trying to get software to work, or debugging it. The experience they want is to go to some website or Github or a tech chat forum, read about something that solves their problem, click the download button, install it and see if it does the job. And if it starts doing the job, they put it in production. They either build applications around it or they install it with their applications so they can monitor their apps. The same thing that happens with Salesforce.com happens with this kind of stuff. Then more people start using it and pretty soon Morgan Stanley has a trading application built on this open source stuff. And then they call the company and say, "Hey, we've got 50 gazillion trades going through your thing here in Hong Kong and we need support for that. Can we pay you $250,000 a year to get priority updates, bug fixes and patches? And it'd be nice if you had dashboards and other things we could use to see what's going on." And that's how it works. That's the floorboard theory. And it's been like that going all the way back to Red Hat? It's generally been the same, but people were getting used to the idea with Red Hat. Red Hat was very broad, so Linux was very broad because you were changing operating systems. Red Hat became the shepherd of Linux in the late ‘90s. What made Red Hat unique was it had influence over the committers, the people actually contributing code to Linux. A lot of them were Red Hat employees. And that's been the pattern that has emerged. Often these start as community projects, but when they get traction and start to look like a real product, it makes sense to wrap a company around it and invest in the company so the company can pay the salaries of the people who are contributing to the source base. It's pretty organic when you think about it. It's pretty organic and it's both socialism and capitalism at its purest. How do you discover open source companies to invest in? Do companies pitch you because of your track record? There are a number of ways. One is me finding them and me deciding they'd make a good open source project. That is not as difficult as it sounds because if you've been a practitioner, you generally have an intuitive feel for what problems people are trying to solve generally, and you have mental boxes. Think of a 9-by-7 grid of various problems that you can put companies into. And visualize check boxes where you say, "OK that area's been disrupted by open source, that one has been, that one has been. But these three boxes haven't been. These three boxes still are closed-source and dominated by Oracle and Microsoft, so something could be done there." And when a project comes along and it's doing that, you look for technical elegance, traction -- whether people like it and use it -- ergonomics, how quick is it, time to value, is it easy to understand, to find, to download and to get it into your app? And because it's all public and social, you can see all of those things happen. Other people see them as well. But what makes me useful to open source entrepreneurs is that there's a lot of pattern-matching and leverage I can bring from all the other open source investments I've done. And also the other thing when it comes down to it: When people talk to me, they realize I'm actually not about capitalism, but I'm actually about software. Does Bain Capital know about this? (Laughs) They do know. I'm probably the nerdiest managing director they have. But you ask a very important question. Part one of your question is how do I know something is good. Part two of your question is how do they know I'm good or useful. That is an important aspect of it. They want a guy on their board, an investor, who understands on a visceral level the value of the software and the value of getting it out to society and just building what people want. And if you build what they want, they want you to survive and they want you to be there to help them, so they want to make it economically viable and they will pay you a few hundred thousand dollars a year to buy the enterprise version of your software. These companies have a gestation period where you have to invest in just building what customers want. You ask them, "What are the top priorities?" And you just build what they want. And get them to build other stuff around your stuff, and get it into production. That's when you're really valuable to them. Where do you find open source start-ups? Everywhere. Open source entrepreneurs are all around the world. Take Zeroturnaround. These guys are in Estonia. I'm looking at a deal in Turkey right now. When I was at Bay, Bain and I co-invested in a company out of Austria. The pattern that works really well is the engineering team and engineering leader, the CTO, are wherever they are. And you build out sales and marketing and other general management either in Boston or Palo Alto. The biggest market for most of these companies is the United States, so it's good to have sales and marketing out here.