W hen we think about innovation, we usually think about a start-up with a big idea, looking for venture capital and anticipating an initial public offering. It is these entrepreneurial companies that attract the attention of politicians, universities, and public R&D centers. And why not? Everybody is hoping for big gains from a small initial investment. They hope the firm will generate the next new thing, build a business, and expand employment in the region that gave them an initial push toward success.

That hope is frequently dashed. Once on their way, innovative companies have a different set of criteria for where to build their business. They look for access to the expertise they need to keep innovating, capital to commercialize and reach export markets, and a labor force with the skills required by their production processes. Innovation is about more than finding and helping talented lone wolves. It is about building industry environments where inventive firms can thrive and grow.

While we have all heard about innovative industries in Silicon Valley and along Route 128, some of the best environments for new advanced technology companies have emerged in unlikely places: our old industrial cities, places like Syracuse and Rochester in New York and Pittsburgh, Pennsylvania. These often-dismissed Rust Belt cities have the assets needed to support process and product innovation, as well as the commercial application of new technologies. If we want to develop a more innovative economy and to translate innovations into products with global markets, we need to find ways to build the nascent industries emerging out of the ashes of our old manufacturing base.

What I call "phoenix industries" have significant features that set them apart from the vaguely defined clusters that are the bread and butter of economic-development gurus. Phoenix industries have what’s called "initial advantage": They benefit from pre-existing personal networks, technical skills, and market knowledge that have developed over a long time, the products of investments in R&D and the workforce made during the heyday of American manufacturing, from the 1950s to the 1970s. Fortunately for our old industrial regions, assets like specialized engineering and research programs housed in nearby universities have proved difficult to move. That is why multinational firms like United Technologies, General Electric, and Kodak have kept R&D operations in Syracuse, Schenectady, and Rochester respectively, even as they have dramatically reduced their assembly line manufacturing there.

Despite their ties to the past, phoenix industries look very different from the old manufacturing industries that they have gradually replaced. Instead of one dominant employer, the sector is made up of many small and medium-size companies. By contrast with their big-firm predecessors, phoenix companies rarely make products that we see on store shelves. Instead, they produce sophisticated components sold to equipment manufacturers, like the high-quality circuit boards certified for use in medical equipment and the defense industry, or sophisticated sensors to measure changes in heat and light used in all kinds of robotic devices. They also design and produce prototypes for products that are then manufactured around the world. They are frequently described as "enabling industries" because they research, develop, and produce technologies that are used in many different industries, instead of just one. And because phoenix-industry companies work closely with a variety of customers, they are constantly engaged in incremental process as well as product innovation.

The story of the photonics industry in Rochester, New York is instructive. Photonics is about the science and technology of light, and it is the basis for a wide range of industrial applications. Rochester, frequently described as a declining Rust Belt city, is nevertheless ranked by the Society for Optics and Photonics as one of the top centers in the world for optics innovation, and the Rochester metro region produces almost six times as many patents per 1,000 workers as the U.S. average. How?

In their heyday, Rochester’s dominant employers–Eastman Kodak, Xerox, and Bausch & Lomb–invested in optics and engineering programs at local universities and built the region’s strength in optics science and manufacturing skills. As a result of these investments, the region developed major research centers in optics engineering and visual science at the University of Rochester and the Rochester Institute of Technology. In the 1980s, these firms pulled out much of their manufacturing in Rochester. But the R&D and engineering programs, rooted in local universities, stayed. Those programs produced a new generation of engineers who foresaw fewer opportunities with a Kodak or a Xerox but more with the growing array of advanced technology firms in the region, the earliest of which began as outsource suppliers to the Big Three. Rochester was also home to thousands of highly trained workers, including quality-control technicians and specialized machinists. Those that could stayed in the region and became the expanding photonics industry workforce. Thanks to this skilled labor pool and knowledge base, more than 100 photonics firms have sprouted in the Rochester region since the 1980s.

Pittsburgh is another example. The city may have long ago lost Big Steel, but it has retained a vibrant industry based on steel-making that exports goods and services to steel producers all over the world. As in Rochester, these small- and medium-sized firms are a diverse group. They produce new types of steel-making equipment, provide engineering services, and do the R&D for devices that improve steel-making productivity and quality. According to Carey Treado at the University of Pittsburgh, an expert on the region’s steel industry, this innovation-oriented industry now comprises more than 330 firms employing 12,000 workers.