Half a century into business, William Miller is still thinking big.

At 82, William Miller might be excused for playing with his grandchildren or simply puttering around. But he's doing no such thing. Instead, he's leading a startup with two very big ambitions: to clean up diesel emissions and to pioneer a new approach to designing chemical catalysts. "I like to be an explorer," he says. "This is an exploration."

Over the course of his 53-year career Miller (not to be confused with a similarly named Cabinet member in the Carter Administration) has been a computer scientist, teacher, software mogul and adviser to venture capitalists and governments. These days he is chairman of Nanostellar, a four-year-old, 22-person firm in Redwood City, Calif. Nanostellar is developing catalysts that promise big improvements over existing ones.

The company's first products are fine powders of gold, platinum and palladium, which, when used to coat a filter for a diesel truck or car, can reduce its toxic emissions by as much as 40% more than existing catalytic converters. At Nanostellar's heart is a computer program that predicts how different compounds will work under specified conditions. Think of it as computer-aided chemistry. Says Pankaj Dhingra, Nanostellar's chief executive, "The impact our tool could have on the world of chemicals is absolutely humongous."

Miller was the last hire of Stanford's Frederick Terman, the "father of Silicon Valley." (Miller joined as a professor at the Stanford Linear Accelerator Center.) He was one of the first five advisers to the venture group Mayfield Fund when it started in 1969. Over the years he also was chairman of Borland Software (nasdaq: BORL - news - people ) and chief executive of sri International, and he advised Chinese government officials, including Zhang Zemin and Li Peng. Colleagues at Stanford (where he was provost from 1971 to 1979) often turned to him for advice. One of those colleagues was professor K.J. Cho.

Cho and a longtime collaborator, Jonathan Woo, had been brainstorming about how to commercialize software they had created to design chemical catalysts. Other researchers use similar tools, including biochemists analyzing protein structures with computer models. Miller's advice: Build a product, not just a tool.

The three went in search of processes that could be significantly improved with better catalysts. Scrubbing the toxic emissions created by the burning of carbon-based fuels was high on their list. But carmakers and utilities have too much invested in existing catalytic converters to adopt such new technology quickly.

Diesel trucks, on the other hand, still used an expensive approach to reducing their emission problems, with catalytic converters that relied on platinum, a highly reactive and very expensive material. The computer model that Cho and Woo developed suggested intriguing alternatives, including catalysts made of nanosize particles of gold and alloys of gold, palladium and platinum.

Since then Miller and Woo have gone on building Nanostellar into a full-fledged company. Miller took the business lead, helping raise $26 million in venture capital from investors that include 3i Technology Partners, Khosla Ventures and Monitor Ventures. He also helped hire the chief executive. "It's more acceptable now to tell company founders they can't run a company," he admits. Woo is chief technology officer.

Nanostellar's catalysts using gold, platinum and palladium are winning technical accolades. Customers are emerging, too, even though Miller ruefully says that business has taken longer to develop than he and his cofounders originally hoped. One European customer is using the company's powder to improve emissions by retrofitting existing vehicles; also, a European carmaker is planning to incorporate Nanostellar's powder in a new fleet of diesel-fueled automobiles.

Miller sees an opportunity to create different catalysts for each class of engine. A ups truck engine, for instance, is different from the one that drives an 18-wheeler. How much time an engine runs "cold" changes its emissions, too, he adds.

He remains upbeat about Nanostellar. Is this his last fling? "Wait until I tell you about my next idea for a startup," he says.

Subscribe to Forbes and Save. Click Here.