Corning Focuses on Customer Data

In the quiet clean of a newly renovated facility in upstate New York farm country, Corning Inc. grows fluoride crystals—highly specialized, glasslike material. Corning’s customers use the crystals to build lenses for the microscopic etching of computer chips.

PDF Download Moore’s Law about ever-faster chips forces Corning’s semiconductor materials group to respond with smaller and finer optics to make them. Fused silica glass, used since the late 1990s to make lenses in chip machines, has just about reached its performance limit. Fluoride crystals are the new thing. Yet two years ago, there was no such business as making fluoride crystals for chip lithography. And five years from now, it will probably be dead, as chipmaking goes on to demand even more precise materials. Nothing stays the same in semiconductor manufacturing.

“We had to get into the fluoride crystals business or we’d be out of business,” says Doug Anderson, chief information officer of Corning’s specialty materials division. “We’ve always got to look at developing the capability to build the next generation.”

But Anderson knew Corning couldn’t succeed in fluoride crystals with the old, homegrown manufacturing systems it had struggled with during the fused-silica era. When it was time to refit plants for crystals in late 2000, the specialty materials group started to devise a technology plan to change the way some key plants operated. At $300 million in sales last year, the division is just 5% of Corning’s total business. But the group has pioneered technology that some of Corning’s 11 other divisions may also adopt, says Rick Beers, corporate director of supply chain at Corning. “We have to keep our focus on what’s being done, not dreamed,” Beers says, adding, “It looks like this is actually working.”

Corning is the granddaddy of the glass trade; the $6.3 billion company has been making and selling it since 1851. The company lost billions last year when the telecommunications market collapsed, driving down demand for the fiber-optic gear on which Corning had made a huge, and ill-timed, bet.

Specialty materials, though, has been a bright spot for Corning, producing low-volume but high-profit products like space-shuttle windows, glass for the Hubble telescope and lens parts for chip machines. But efficiency hasn’t been the division’s forte. The nine specialty materials plants all ran incompatible applications, manually coded in Microsoft’s Visual Basic. And the fact that each factory had its own profit-and-loss responsibilities meant they didn’t want to share data anyway. Plant managers were more focused on hitting their own performance goals than helping out other factories. “There was minimal sharing of information,” Anderson says.

If one factory was overloaded with work, unfinished glass couldn’t be finished at another site because the two computer systems couldn’t communicate. Production would get backed up, leaving key chip-machine customers such as Nikon and Canon waiting. Sometimes, if the order was worth it, Corning would buy an extra machine for the congested factory even if an identical machine was sitting idle somewhere else—an unplanned expenditure of $800,000 or more. And when a factory found a better way to do something, the breakthrough usually couldn’t be duplicated at other plants because the information systems were so different.