From the Factory Floor

A monthly column of assorted news, interviews, and technical information from Metrowerks.

By Dave Mark

This month we’ll meet Metrowerks’ colorful Chairman of the Board, Jean Belanger. Jean hails from Ottawa, Canada, but now calls Austin, Texas home. He got a BComm. from the University of Ottawa and an M.S. in Finance from the London School of Economics, and is also a Canadian CPA. Prior to Metrowerks, Jean was a venture capitalist and an investment banker.

Dave: When did you first meet Greg Galanos?

Jean: I first met Greg in 1989 when he was looking for capital for his company, Metropolis Computer Networks, Inc. At the time, Greg was selling Modula-2 for Macintosh and I was running a C$9 million technology start-up fund. Greg was utterly relentless and finally I invested C$100,000 in his company so he would leave me alone [laughter]. It was the last deal the fund undertook as we were fully vested.

Dave: What was your role in the company?

Jean: In 1989, I became Chairman and we started re-making Metropolis into a company which could get to the next level. We changed the name of the company to Metrowerks Inc. (for trademark reasons), and we went about finding new business opportunities. In late 1989, Metrowerks developed under contract a Modula-2 front-end for the MIPS RISC-based microprocessor. This gave the company an early view into RISC-based computing. Subsequently, the company developed Pascal for Macintosh and started a project to develop Pascal/Modula front-ends and code generators for Sun Microsystems’ SPARC microprocessor.

In 1990, I decided to leave the venture fund to help re-structure a mainframe software company. In May, 1991, Greg asked me to come in and help him decide what he should do with Metrowerks. Our first job was to look at Metrowerks’ business from a strategic point of view. It became apparent very early on in our strategic review that Metrowerks would have to jump into the core C/C++/Pascal tools market if it ever hoped to become a self-sustaining business. We thought Metrowerks could make the transition, for three reasons:

1. Metrowerks had its own IDE, while most of the other programming tools vendors except Symantec relied on MPW. These other vendors did not control their own destiny. The Metrowerks IDE was not great, but it was a major achievement for such a small company.

2. The shift to PowerPC would force everyone in 1994 to buy new tools. Therefore, we guessed there would be a 12-month window of opportunity.

3. We would be building a new RISC-based product from the ground up. We did not inherit a lot of legacy code which needed to be bolted into a RISC product. We had very few customers at the time and we could ignore what was going on around us and focus 150% on CodeWarrior.

So in July, 1991, we decided to attack the mainstream Macintosh tools market for C/C++/Pascal. It took a little under a year to put the preliminary product plans in place and for the first fund-raising to be completed. In July, 1992, we stopped all development on all existing Metrowerks products to focus 100% of our technical and financial resources on CodeWarrior and its product launch which was penciled in for early 1994. It was full steam ahead.

Dave: What was your original goal as a company?

Jean: We had two major strategic goals and everything we have done since 1991 is based on them:

1. We would build the world’s best programming tool no matter what it cost. I committed early on to raise whatever capital Greg needed to make CodeWarrior number one in the market. So far that has meant raising over US$8 million.

2. We would change the “agenda”. What I mean by this is that we really worked hard to figure out what Mac programmers needed. Among the major decisions we made were:

• Multiple front-ends - We decided from day one that we would do a native PPC Pascal. Everyone was abandoning Pascal and we thought this was a dumb thing to do.

• Common front-ends - We decided early on to have our 68K and PPC code generators host the same language front-ends. This was one of the keys to making porting to PPC easier: move existing 68K code to CodeWarrior 68K, and then to CodeWarrior PPC. This is also the key for us getting Java to market.

• Other Processors - Our common front-ends give us huge technological leverage to attack other processors, as they reduce the amount of work required to support a new processor code generator by at least 50%. Our front-end tokenizes and compiles source code into a processor-independent, intermediate representation. The back-end code generators take this processor-independent representation and turn it into object code. The front-end cares about things like the difference between a while and a for loop, while the back end cares about things like PPC instruction pipelining. The point is, when we add support for a new processor, we only have to rewrite the back end. The front end just gets recompiled (with a little retooling, if necessary), since it is processor-independent.

Our view is that the more platforms we support, the more valuable the skills of Mac programmers become. In other words, if one IDE can build Mac apps, Windows apps, PDA apps, etc., the greater the demand would be for Mac-based programming skills.

• Core Tools - We decided, after very heated internal discussion over a two-year period, that we would not get into the database front-end application development business in a big way. This meant we would not try to build a cross-platform framework. This allowed us to really push PowerPlant to be Mac-specific and the most advanced framework available on any platform.

Dave: Was Metrowerks always a Macintosh company?

Jean: Metrowerks was a Macintosh company from the get-go and this has not changed. We recently announced that we would host our IDE on Windows, but we are doing this via the Altura porting libraries. What this means is that all IDE work at Metrowerks is first done on Macintosh and then ported over to Windows95. While we are very much a “Mac” company, our hope is that we will become the tools vendor of choice for all Mac development and for all embedded systems development. By embedded systems we mean all non-Mac and non-Windows development. While everyone in the business seems to have Windows on the brain, we would be happy to have the rest of the business [laughter]. Why can’t the Mac be the number one platform of choice for developing embedded systems applications?

Dave: Did Apple welcome you with open arms? Did they see you as saviours? Did they see the value in your creating a PowerPC native development environment? Did they believe in you?

Jean: No large vendor, Apple, IBM etc., likes depending on a small vendor. In January, 1994, when we launched CodeWarrior DR/1, a lot of Apple people thought we would be a short-term fix. We got a lot of help from Jordan Mattson early on. He basically risked his career several times by “directing” to Metrowerks Power Macintosh prototypes at critical times, etc. Jordan was in a difficult position as he works in the Apple Developer Tools Group, which is funding development of Symantec C++ and MrC. On the engineering side, Apple engineers were fantastic and really helped us by running their code through our compilers.

So in the beginning, we were a temporary “fix” to the problem until more “substantial” players came along. This was never our view. Developing CodeWarrior was a brutal three-year grind and we have no intention of giving way to any vendor. Currently, we have over 45 engineers working on CodeWarrior, not including technical support. This constitutes one of the largest developer teams in the world working on a single Mac product. We have completed eight releases in 24 months. We are now the number one selling programming tool in the world for Macintosh development. We have over 30,000 registered users. We now have over 75 employees and are based in Austin, Texas. We believe we have earned our place in the Macintosh developer community. We have come a long way in a relatively short period of time.

Dave: A lot of our readers are interested in the start-up process. Can you describe the process you went through to fund the company?

Jean: Most people do not understand the money business. The first thing anyone raising money needs to understand is that money is a commodity. Money flows to those uses which generate the highest rates of return given the risk profile of the “holder” of the funds.

What this means is that you have to do three things when raising money:

1. You need to convince the holder of the cash that you will generate “x%” rate of return.

2. You need to convince the holder of cash what the risk profile of your venture is.

3. Never leave yourself only one “type” of funding source.

Before you waste your time trying to convince someone to give you money, you need to figure out what their risk/reward profile is. Don’t waste your time trying to convince my grandmother to invest in your software project; she is not interested. She holds “blue chip” stocks for dividend purposes because this represents the biggest portion of her “personal income”, which she needs to live on. She will never risk her “personal income”, no matter how great the potential reward is.

Too often, entrepreneurs think they have all the answers. The only answer that counts is if you win. There is no prize for second place. If you don’t get the money, your business dies. So don’t waste your time trying to sell someone who is not buying. If you find players who are willing to undertake the level of “risk” you are selling, then focus all your energies on these people. I wrote seven business plans for Metrowerks in 24 months. Each one was tailored to the audience I was addressing - which leads to that third point. Never ever leave yourself open to one type of funding killing your dreams.

In other words, don’t just go after venture capital, or supplier financing, etc. Make sure you have a lot of avenues open and that you are really working them hard. One of the most important pieces of our funding in 1992 came from a book publisher! Another came from a friend, Stephen Lockyer, in Halifax. Metrowerks was funded by private placements, R&D tax credits, friends, you name it, we tried it. We were not too proud to take anyone’s money because we might have thought that it was not the right “flavor”. Remember, it’s all “green”.

In almost all cases no one wants to be first in a financing. If you have multiple sources going at one time then you can leverage them. The book publisher who gave us a small amount of money to do a Pascal higher education product was used as evidence that we could market our products, which led to an R&D loan, which we repaid from our IPO, etc., etc.

Raising money for a capital-intensive software project like CodeWarrior is a full-time job. Greg was spending the money on CodeWarrior as fast as I could raise it. We both went several months without pay to keep R&D fully funded and on track. I trusted Greg to deliver the product. He had to trust me to deliver the money. It was pretty hairy at times but we had the necessary skills in-house to get this done. We did not depend on consultants etc. to get us the funding. Greg and I each owned 45% of Metrowerks when we did the first round of CodeWarrior financing, so we could both speak to our respective constituencies with authority.

Raising money is a nightmare. If you can’t live with months of bad dreams then don’t start a software company, which requires a lot of capital. The upside is that there is a lot of money available. The problem is that you have to go find it. The second problem is that there are no excuses: you either get it or you don’t. This makes it tough.

Dave: Where do you see Metrowerks going from here?

Jean: We have a simple frame of reference. We want to double revenue every year. This is the view from twenty thousand feet. In the fiscal year ended July, 1995, we did US$5 million in revenue. This means for the fiscal year ended July, 1996, we would need to do US$10 million, and so on. I am not saying we will do it, but this helps guide us as to where we are going.

Revenue drives share prices and in software you need to keep increasing your share price. Why? Because if you don’t you will lose all your key technical people to competitors who can hold out the promise of higher prices and valuable options. That is not to say that all engineers look to is options. But you are crazy if you think engineers can’t count and keep score. So if you keep making “financial” progress, you keep your people. A simple rule but it works.

We look at markets 24-36 months out and figure out how much revenue potential there is so we can double every revenue when we get there. We try and make sure we have a lot of overlap so we can lose on some and still make our target. For instance, we are getting ready to launch Java products in May of 1996 for Macintosh and for Win95 in the last half of 1996. How is this going to impact our sales? What is the overlap?

We see our growth coming from the following markets:

Our Core Mac Business - This continues to grow. The notion that Mac developers are leaving the platform just doesn’t hold up when we look at our numbers. There seems to be growth right across the board from where we thought we would be at this time. I can’t believe all the bad press the Macintosh market gets when we look at the cool things our customers are doing developing apps for the Web, the internet in general, multimedia authoring, technical subjects, etc.

Java/JavaScript - Java and the internet are great for us, because the success of the Web is going to make the world more C-intensive than ever before, and Web content authoring is going to include more programming “resources”.

Java was derived from C++, and C++ is an extension of C. It is really hard to “stuff” a big, fat 4GL binary down a small thin “network” pipe. What everyone forgets is that the programming tools market is only so big. If a potential customer chooses to use a 4GL or other related tool, we lose revenue. The internet is changing the needs of users. If every company on the planet ends up with a Web site, there will be a lot of content written in Java (or related languages) and a lot of CGIs written, etc. These can’t be done with traditional tools. This is great for Mac programmers, given that recent surveys show that Web content is most often authored on the Mac. Web content providers will have to increase their budgets for programming resources to get all that cool Java stuff up and running. Where will these budgets come from? Traditional advertising budgets, you name it. No one wants to miss the internet action.

Embedded Systems - We are pushing real hard to open up this market for desktop programmers. For instance, we would love to support all of the major games vendors. Wouldn’t it be great if you could use CodeWarrior to program for all the games platforms, all from the same IDE and on the same platform? This is just one example of the synergy of generating code for several processors in one product.

We are breaking new ground in the embedded market, as this has traditionally been a very Unix-centric market. We believe that CodeWarrior at US$400 per copy running on a PPC604 Mac is a killer development platform. We have been selling this concept for the past three years to the big systems and microprocessor vendors. This is where the big push came from for us to host CodeWarrior on Windows. All of the big embedded players want to offer their prospective ISVs a choice of using either the Mac or Windows to program their “systems”.

In summary, I see Metrowerks growing for quite some time to come. I really believe we have one of the best R&D teams on the Mac and as a result we have a killer product. The wild card for us in all of this is if Apple increases its market share in the markets we are interested in. While it seems everyone is negative on Apple right now, I think things will turn around. I don’t think the world is ready for only one desktop API.

Dave: What’s next for you?

Jean: We have hired a great CFO in David Perkins, so I am spending almost all of my time on revenue generation and marketing. In 1996, I will be spending a lot of time on Java/internet-related issues given our R&D relationship with Sun Microsystems.