NEW YORK (FORTUNE) - Aart de Geus is one of the most insightful people I know in the technology industry, so I was happy to sit down for lunch with him this week. De Geus is CEO and co-founder of Synopsys (Research), a company that sits at the intersection of just about every trend there is.

That's because it is one of the two main makers of the software used to design most of the world's most cutting-edge semiconductors. Chips are so complex today that figuring out how to lay out their intricate networks of microscopic circuits is one of the great achievements on which all industry -- and social -- progress today relies.

More from Fortune 10 dumbest iPhone apps The surfin' CEO McDonald's wins with global palates FORTUNE 500 Current Issue Subscribe to Fortune

Michael Dell on Alienware, growth, and AMD Dell bought Alienware. Alienware uses AMD. Will Dell itself follow? (Read the column) More Fast Forward The new Net boom Don't stare so hard at Google you miss Yahoo! IBM's Quasar: Is it the future of computing? Owning up to global warming Fast Forward archive

For de Geus, just about the most important trend in tech -- and for his company -- is the movement towards mobility. He says the cellphone has taken on the mantle held for decades by the PC as the device that drives the most sophisticated innovation.

As companies like Nokia, Motorola, Samsung and even, if rumors are to be believed, Apple, work to incorporate voice, MP3 players, cameras and video into smaller and cooler devices, the chips inside will need to be correspondingly more capable and complicated.

"There's a hunger for integration on phones and PDAs because of all those functions," de Geus says. He is certain we're heading toward one single convenient multi-purpose device -- a sort of portable miniaturized version of the PC.

"You don't want a three-year-old iPod or phone," he continues. "You want this year's model." He might have said "this month's." But he rightly notes that there isn't the same hunger for evolution on the PC.

"On your computer today there's probably nothing you do that's not fast enough, except turning it on and off," he says. "But the cellphone has many years ahead of it of getting better and better."

The ultimate driver in all this for the chip industry -- and thus for Synopsys' sales -- is ever-improving digital video. "That means umpteen-bazillion bits have to be processed, which means unending investment in tech."

Synopsys is number two behind Cadence Design Systems (Research) in what's called Electronic Design Automation, or EDA. (Cadence's revenues last year were $1.3 billion, while Synopsys' were $990 million.)

The traditional mainstream of EDA includes two businesses -- software that designs the circuits on chips, and the separate software that manages the testing and verification that the chip designs actually work. The two companies sell to almost all the same chip-making customers, who are generally loathe to rely on any one supplier for such critically important products.

There is another newer business in which Synopsys and Cadence also compete called "Design for Manufacturing." (It has its own DFM acronym, as seems obligatory in anything connected to chips.) DFM uses something else called Technology Computer Aided Design (TCAD of course), which enables three-dimensional visualization of individual transistors in a chip design.

Every "fab," or chip manufacturing plant, has its own recipe for making transistors. At the infinitesimal scales at which chips are now made, a design that worked in one fab might not work in another, because of atomic-level peculiarities of the process. A company would use DFM software to study and adapt for one fab's transistor designs and then, as de Geus says, "use that information in the chip design where you are using 79 million of those transistors."

So, for example, a so-called "foundry" company like Taiwan Semiconductor, which fabricates chips, might share its unique transistor characteristics with nVidia, a "fabless" chip company that designs chips that incorporate tens of millions of those transistors (using Synopsys software).

Finally, Synopsys has a fourth business that has taken off in the last couple years. As chips get more and more complex, a market has developed for pre-designed components. So Synopsys sells complete modular chip architectures for things like USB connections.

This intellectual property business takes EDA in a totally new direction, but de Geus says it's a natural consequence of the industry's evolution. "When any high tech field matures, innovation moves from point innovation to systemic innovation," he says.

So Synopsys is now a company that builds the basic software for chip design, goes down into the atomic guts of the silicon, and takes complete designed components into systems.

The industry, and Synopsys, had a bad year in 2004. Its stock dropped in half, partly because of a move toward selling more software on a subscription model, which significantly reduced revenues, since many sales were formerly recognized up front.

But the company's sophisticated at-the-crossroads formula seems to be working now. The stock is up over 20 percent in the last year. To learn more, visit the web on your cellphone.

Fast Forward is a weekly column by FORTUNE's David Kirkpatrick. He can be reached at dkirkpatrick@fortunemail.com.