John McTague: I'm here as a token representative of the Rust Belt industry. You might ask what somebody from an automobile company is doing on a panel discussing entrepreneurship. I'll try justify that. It may be hard, but I'll try.

The automobile industry is a, it's almost a one-of-a-kind industry in its scale if nothing else, scale and complexity of product. It's a very huge industry. We heard yesterday about these hyperbolic projections that Internet commerce in the year 2010 or so was going to be on the order of $350 billion. Well, that is roughly the revenue of just two automobile companies that happen to be located in Detroit. And between them, they only have about 28 percent of the world market. So it's a big industry. It's an industry which, apparently unlike what we've been talking about here, has enormous economies of scale. The proof of that is that early in this century, there might have been a thousand or so automobile companies. Today if you look at the ones that are significant, there are, as of recently, two American companies, three German companies, zero British companies, three Japanese companies, and one Italian company. And all of the rest are essentially in the noise level. So by looking at what has actually happened, it's clear that there is more and more a pressure toward economy of scale, and everyone believes that that is going to continue and that more and more of the marginal and even sort of average companies will be swallowed up by the more effective large companies.

Where's the economy of scale come from? A lot of people will say, well, it's the plants. Gotta have a very big plant. There's a huge capital investment; you gotta get a lot of vehicles out of a single plant. Well, it turns out that the optimal size of a plant is one that produces around 200,000 vehicles a year. Now, that may sound large, but the world production in a given year is around 70 million vehicles, which means that the optimal, roughly speaking, the optimal world number of plants would be around 350. So it's not as though there is a pressure to go to 10 plants or 20 plants or something like that. The economies of scale are not in the plants, and in fact there are an excess number of plants in the world, substantial excess, maybe enough to produce 90 million vehicles instead of the market of 70 million.

Where is the economy of scale? It turns out the economy of scale is in product development, how you actually do the engineering that produces the vehicles. If you look at the R & D budgets, the two companies with the largest R & D budgets are also located in Detroit; they're the same two companies. General Motor's or Ford's R & D budgets, each of them are somewhere around $7 billion a year. And the business involves making new vehicles every year. The scale of making a new vehicle is in the billions of dollars, the R & D costs. It's like being in a business where you have to produce two 777's a year in any one company. So it's a very R & D intensive industry.

Now in the old days, as you know, the way vehicles were made were a draftsman drafted the shape of a part on his drafting board - and it was a he, that somebody - then a machinist tried to make the part. He assembled all the parts in a room and tried to stick them together, and if you banged a little here and banged a little there, usually they'd stick together. If they didn't you'd shave a part down a little bit. And then if that worked then you'd go to the people running the plant and say, "Make these parts for us." And they'd say, "We can't make those parts; you've got to shape them this way. Our machine won't do that." So then you'd go back and, eventually, after several years, there would be a design of a vehicle which could be manufactured. And the time scale, say, fifteen years ago, might have been about six years to design a vehicle, because they're so complex, and you had so many iterations including, by that time, safety tests. You'd do a crash of a vehicle, find out that it wasn't acceptably safe, then you'd have to redesign the vehicle, build another prototype, do it again, do it again, and each one of those cycles would take six or eight months at very substantial costs, and that was one of the reasons why it took so long to produce a vehicle. And it also was one of the reasons why you didn't try too many different variants, because it was extremely expensive and time-consuming.

Well, the way product development is going now has to do with really what an engineer does today, which is very different. Engineers don't sit down and draw things with pencil and paper. They don't even just sit down there and look at a CAD screen and do the equivalent of that thing on a computer instead of using a drafting (table) as they did, probably, 20 years ago.

The clue to the value added in product development today isn't the product itself, by the way, because no matter what you do, somebody else is going to do it within 18 months. It's the way you do the development. So there's been an enormous change in the way development is done, moving very much to design based on fundamental analysis of the properties of materials, of the properties of combustion, etc. Using that analysis, doing extremely sophisticated simulation of systems and of processes for manufacturing vehicles for logistics and, as we heard yesterday, how do you link suppliers into this whole thing?

As was mentioned yesterday, Ford, General Motors and Chrysler are getting together to develop information-transfer methods such that we can do simultaneous distributed design with our suppliers, many many hundreds, thousands, of suppliers, in a way that is secure from one company to the other, yet it is relatively transparent to each of the suppliers so that they don't have to have three different design sets of hardware and software. They don't have to worry who they're designing for, so to speak, but we can do work simultaneously and interactively with our suppliers, without getting cross-talk from General Motors or Chrysler.

So, nowadays, for example, instead of doing this building a prototype of a vehicle, crashing it into a wall, looking at the pieces and deciding, well, this would have involved a certain level of injury to a driver or a passenger, we do the crash tests on the computer. The computer model includes basically everything, even the positions and strengths of all the welds in the vehicle, and a full crash test takes about six to eight hours on a big crate. The results that you get out are more reliable than what you get from a physical crash test. And it has also the clear advantage that you can do extremely many options. You can change the position of a weld, see what that does. You can change the position of the steering wheel. You can change the material that the steering wheel is made from. You name it. You can do all these changes. Change the interior configuration of the vehicle, whatever you wish to change. So you actually, in this product development and manufacturing system, where in fact there are huge economies of scale and smaller numbers of producers, the number of options that you can examine has increased substantially. And one of the net results of that is that there is a large increase in diversity in the types of products that are out there today. You don't just have four-door sedans and two-door sedans and stations wagons. You have all of those things, you have SUVs, you have ATVs, you have little pickups, big pickups, two-door, four-door, you have mini-vans, you have things that are hybrids between all of these things. You have even alternative power plants.

William Janeway: And unfortunately we are beginning to run out of time.

John McTague: So we have these tremendous variations that are occurring today. One of the net results of this, by the way, is the shift in the type of person who does the engineering. The people who do engineering now need rather substantially higher levels of education; one-third of all the engineers we hire today have master's degrees. The number of Ph.D.s in the company has doubled in five years. As I mentioned, the product-design cycle has dropped by a factor of three, and the diversity that's available to the customer has increased substantially. And this has all been a result of the way information can be transferred and the way design can be done in a distributed way, indistinguishably, whether it's done in our house, in suppliers', done in collaboration of an engineer in the United States with one in Germany or Britain or Australia, so it's a remarkably changing industry because of the information age. Thank you.