Warren told us that the semiconductor industry, after a promising youth, is now "middle-aged", but still young at heart. Since the transistor was invented almost 60 years ago, developments have tended to come in waves. The current wave is bulk CMOS technology, which has had a very successful run. Over the years it has been drastically shrunk in size, so that it needed less and less power, and cost similarly less to make. This has led to a virtuous circle of more sales, more investment and hence continued improvement in performance. But problems are appearing. Current R and D is working on 32 nm features, but at this size transistors do not always behave as their models say they should. So R and D costs are rising, but customers expect prices to go on falling, while shareholders, as ever, expect returns.
The ARM company, of which the speaker is Chief Executive Officer, had for its parents two computer companies, Acorn, which produced the extremely successful "BBC" micro-computer, and Apple. Acorn had followed up the BBC micro by developing a 32-bit RISC (Reduced Instruction-Set Computer) in 1985, but subsequent need for finance led to the formation of ARM in order to exploit it. This RISC design has been around, in successive implementations, for 21 years, and now appears in some 10% by value of all the world's digital ICs.
But ARM does not actually
ARM started in about 1990 with 12 engineers and one salesman, setting themselves the cheeky challenge "Can we sell a microprocessor to Intel?" It now has about 1200 engineers, and 1300 employees altogether, but claims that the end users of their designs number 1.2 billion people! Their policy is to license the technology to many different "partners", who then make chips which the equipment manufacturers put into a great variety of products, from mobile phones to anti-lock brakes. The partners then pay for it mainly by royalties on the chips they sell. They have negotiated about 400 licences so far, of which about half are bringing in income, thus funding further development, and paying dividends to shareholders.
For management, the problem is how to encourage innovation. For instance, the reduced size of modern chips mean that the transistors become more variable, and don't last as long either. So errors occur in the logic, as they do too if one reduces the power supply voltage to economise on power consumption. But with increased complexity, and clever engineering, errors can be corrected, provided there aren't too many of them.
In fact, it is in times of crisis, towards the end of a project, that people are often most innovative. But because it is a crisis, there is often no time to use the new idea. So there has to be an arrangement for "banking" it, for use some other day. There needs to be a mechanism for giving people "dead time" between projects, and sometimes for rotating them through research groups. You have to take risks, and recognise that people will make mistakes.
The speaker continued with some observations on University/Business collaboration. Universities supply the graduates whom industry employs, and its need is for people who are not only technically competent, but can also appreciate business attitudes. Academics were now showing a tendency to move into business themselves, but there were right and wrong ways to do this. The "spin-off company" is showing signs of being a right way. Business and academia operate in "parallel universes" because of their different time-scales, but the universities do perform the essential function of doing the underlying science. Electronic firms have always been very bad at doing this for themselves - pharmaceutical ones have done much better.
The lecture concluded with a few more brief observations:
Be economical - don't over-engineer things;
One often comes up with a better product the second time round;
Technology and business are inseparable;
The commercial world is not "dirty", but essential;
Keep it simple, simple, simple;
Innovation relies on a culture that encourages it, the necessary basic science, and an economic climate that allows business to flourish.
|<< Previous article
|Next article >>
|SOUE News Home
Copyright © 2006 Society of Oxford University Engineers