I think there are a couple of points, that others have alluded to as well, that show why the automative market is a better bet than some might think. One is that upgrading the electronics in a car can be done at any time, and it can be installed in any number of different models produced by a manufacturer. Think of systems like OnStar, which are installed in all GMs, and which can and are upgraded irrespective of other model changes. Or entertainment systems in cars, which also can change at any time.
The other thing is that all of these electronics, as well as power steering systems, transmissions, engines, batteries, climate control systems, windshield washers and wipers, brakes, shock absorbers, wheels and tires, and on and on and on, are certainly modular and are installed in many different models. That's why anyone who looks under the hood or underneath, can see the strong family ties between, say, a Chevy and a Cadillac.
And there's another point. Cell phones and tablets are basically toys. Toys never last, because people get bored with toys and muct have a different one before long. Cars are more like, you know, your kitchen stove, your washing machine, even your house. These aren't things you change at the drop of a hat. And yet, compared to those other more or less permanent fixtures, cars do evolve quite rapidly.
@Junko: Carmakers say they need innovation, and yet the typical automotive product development cycle is about five years. In most other industries, the whole world -- of technology, market trends, consumer preferences, and pricing -- changes within three years.
It depends on what you mean by innovation, and points out fundamental differences between the consumer electronics and automotive markets.
Auto design cycles are longer than consumer electronics design cycles, and pretty much have to be becasue of the nature of the product. Retooling a factory that makes cars is a fundamentally different exercise than combining new chips on a motherboard.
Automaking innovation will perforce be incremental, and the process will be marked by lots of little changes, not big fundamental ones.
I'm old enough to recall when cars weren't a market for chip makers, because they didn't use chips. Everything was mechanical linkages. Now any car you buy witll have a plethora of sensors and microcontrollers, communicating over an internal local area network with its own specialized protocols.
This did not happen in a sudden dramatic matter. It happened slowly, step by step, with each step of the way marked by the questions "What will it cost to do it?", "What will we get by doing it?", and (most important) "Will the market go for it?" If you are an automaker, you want to limit your risks, and not do things that will kill you if they aren't successful.
I think if you could tour any major automaker's R&D labds, you would find engineers working on all sorts of things that are innovative, but they might not be things you would even be aware of if you buy a car that incorporates them. They will be designed to reduce costs, improve servicabilty, and make cars safer to drive.
To top it off, automotive doesn't exactly offer either the fastest growing or the largest volume market for semiconductor companies.
No, but they offer a stable and steadily growing market large enough to be served profitably if you can get design wins. Name me a chip maker that doesn't produce a variety of "bread and butter" chips that may not get the volumes of being selected for the new iPhone, but do provide steady revenue and profits that keep you afloat if you don't get the Apple design win.
I have often wondered how far you could take building up a car by plugging together major components built to standards, just like we can do with PC boxes.
I might argue that auto manufacturers increasingly are doing that, to achieve lower costs and economies of scale. IIRC, for example, Chevrolet actually makes the engines for most of GM's non-truck lines. In many cases, models will use the same chassis and running gear, with only the body differing between models.
I haven't Looked Stuff Up to get details, but my impression is that the industry has been steadily moving in that direction for decades.
In many respects, automakers mirror the PC market. Outside of luxury brands, cars are commodities, margins are thin, brand loyalty is fleeting, and lowest cost producer is likely to win.
The fundamental mismatch is that cars last too long for the product cycles in the electronics industry. We did some consulting to Visteon years ago, and the advice that I gave them then was to standardize interfaces and form factors in the cars for electronic components. I have often wondered how far you could take building up a car by plugging together major components built to standards, just like we can do with PC boxes. At a minimum, there should be standardized slots for entertainment and navigation systems with defined power capabilities and limited access to automotive data. This is more than the current situation, where there is sort of a standard-sized slot in the dash and a very customized wiring harness.
Unfortunately, this advice is fundamentally counter to the business model of automotive manufacturers. Their insistance on building cars as vertically integrated platforms is great for them, but it severely limits customers to what they offer unless the customers are willing to hack into their cars.