The electronic design-automation business has proven in the past to be a nearly flawless predictor of where we are headed. The bad news first: Times are tough, and that will not change much throughout the year. I expect 2002 to be a near-zero-growth year for EDA.

The good news is that EDA is critical for the coming system-on-chip economy. We can expect that EDA, as the core enabling technology of supercomplex systems-on-chip, will experience momentum and growth before any other part of the electronics industry. In fact, RBC Capital Markets characterizes EDA as one of the must-have technologies for semiconductor designers.

The next positive economic cycle will depend more than ever on new silicon chips. Without new power tools, there will be no new superchips. Without enough new superchips, there will be fewer exciting new consumer products to buy.

It has been tough to talk about purchasing new tools and adopting new design methodology at a time when chip companies are downsizing their engineering organizations and are undergoing severe budget cuts on an almost-regular basis. Nonetheless, the winners will be those that find a way to make the investment during this cold "nuclear winter."

Design starts, the ultimate measurement of electronics industry momentum, are at an all-time low. In the first half of 2002, the industry initiated a third of the starts routinely accomplished over the same period in previous years. To come out of this down cycle, we need to see the number of design starts grow significantly. What will be the driver?

The electronics industry at large has been slow to introduce new products and services. Yes, we saw a colorful new personal computer (the iMac), and mobile phones are getting smaller. But with next-generation systems-on-chip, it will be possible to create products and services that will be out of this world and centered on bandwidth and convenience.

At the same time, we continue to use electronic gadgets (mobile phones, PDAs, Game Boys) that were designed years ago. Most chips that are in volume production are well below 10 million transistors. But with new design starts targeted for 130- and 90-nanometer processes, I expect average chip design complexity to increase drastically, to 20 million transistors, within the next 12 months and to 200 million transistors by 2008.

Before the 2001 bubble burst, insiders talked about massive retooling for SoC design and development. That has not happened yet, but it will. And it will be brutal in terms of complexity and natural selection.