September 04, 2007
By: by BusinessWeek
Lessons from IBM's innovation factory
How IBM makes radical collaboration work.
By late 2003, IBM's decision three years earlier to pump $5 billion into its chip business wasn't looking so smart. The division had lost more than $1 billion in 2002 and was on its way to losing $252 million more in 2003. Investors urged Big Blue to quit, but that wasn't going to happen. IBM saw leading-edge chip technology as vital to keeping its lead in the highly profitable business of making powerful server computers. Still, clearly, something had to be done.
That's why John Kelly, who then ran the semiconductor division, summoned 10 executives to IBM's chip factory in East Fishkill, N.Y. Kelly argued it was time for a new strategy. IBM needed to share its most advanced semiconductor research with a few key allies. The tech giant already had a handful of alliances aimed at improving manufacturing and chip design. Several partners had come forward asking for deeper relationships, including collaboration with scientists working for IBM Research. Kelly expected pushback from his people, and he got it. "It was a real struggle. We had never thrown our doors open before," recalls
Bernard Meyerson, who then ran research and development for the chip division and now manages alliances. "We could all envision nightmare scenarios of a decade of research value being lost."
After two hours of heated debate, Kelly prevailed. Since then, IBM has built what it calls an "open ecosystem" of chip R&D with nine partners, including Advanced Micro Devices, Sony, Toshiba, Freescale Semiconductor, and Albany Nanontech, a university research center. All told, in five separate alliances, IBM partners have contributed more than $1 billion to help expand the company's facilities and buy the latest chipmaking equipment. But just as important, they're providing brainpower, including more than 250 scientists and engineers who now work in East Fishkill. As a result, IBM's chip operation boomed, and, even now, during a cyclical downturn in the chip industry, it's still making a profit.
IBM is reinventing the way it innovates. At one time the tech giant was a true believer in go-it-alone R&D. The feeling was that if a technology wasn't invented by IBMers, it wasn't as good. Now the computer pioneer realizes that no matter how big an organization is, more smart people are going to work outside its walls than inside. So it courts R&D partners aggressively. "We are the most innovative when we collaborate," declares Chief Executive Samuel J. Palmisano.
In an era of fierce competition, it pays to innovate communally.
IBM's decision to invite in outsiders and open up the innovation process reflects one of the most intriguing concepts in corporate strategy today. Many major companies have concluded that succeeding in the 21st century requires teaming up with other companies-or even individual researchers-to create so-called innovation networks. "These networks allow companies to seamlessly weave internal and external innovation capabilities to optimize profits and speed products to market," says Navi Radjou, an analyst with Forrester Research.
Companies no longer compete simply against one another. Now alliances devoted to innovation go head-to-head. bt Group, basf, Boeing, Eli Lilly, Procter & Gamble, and IBM are the pioneers. They all have revamped their strategies to expand collaboration with outsiders. Forrester estimates that while most major companies are aware of innovation networking, only about 20% to 30% are experimenting with it, and a mere 5% have mastered the practice.
Forget the cookie cutter. Get out the chainsaw.
There's no one-size-fits-all approach to collaborative innovation. What works best overall, strategy consultants say, is to think radically.
Some companies turn suppliers of goods and services into something much more valuable-sources of ideas about how to design a product and its components. Boeing, for instance, tapped a global network of suppliers to produce much of the detailed design work for its new 787 Dreamliner jet.
Other companies are busy prospecting for valuable new ideas from individuals and startups. Britain's bt placed scouts in India, China, and Silicon Valley to spot useful inventions and funnel them into its businesses.
Yet another approach: Bring together a handful of companies to sharpen their competitive edge together. Businesses have been doing that sort of thing for more than two decades, but the alliances are different now. Increasingly, they are global, not national. In 1987, hard-pressed U.S. chip companies set up a research consortium, sematech, to counter Japan's growing power in the chip industry. Today, sematech has two Japanese members. Another change: Scientists from different companies work more closely than ever, sharing ideas and intellectual property in ways their research forefathers would never have considered.
Some early results are in. Boeing cut 12 months off the time it took to bring the 787 Dreamliner to market. P&G improved its R&D productivity by 60% through its Connect + Develop program, where it links up with individual inventors from around the world. And at bt, teaming up with outsiders on innovation projects has spawned $1 billion in incremental sales for the telecom giant since 2002.
Rust Belt rebirth: IBM's factory of the future.
Seven years ago, IBM's chip fabrication plant in New York's Hudson Valley was in mothballs. Today it's humming with activity. That's thanks in no small measure to the company's alliance strategy.
The 200,000-square-foot factory cost $4.4 billion to rehabilitate and expand. IBM shared the huge cost with its partners. "By ourselves, we'd have to bear expenses that are just enormous-and well beyond our appetite," says William Zeitler, general manager of the company's Systems & Technologies Group.
The shared investments buy a load of cutting-edge technology. In the vast plant, you feel like you have crossed into a world where robots are in charge. There are hundreds of giant chipmaking machines as large as locomotives. An automated overhead transport system carries stacks of silicon disks in picnic-cooler-size pods. The pods, lowered on cables, feed the disks into machines that will move them through each step of the assembly process. Within the machines, robotic hands pass the disks from one workstation to the next, where the surfaces are etched, coated with chemicals, or baked. Each disk typically holds 500 to 1,000 chips.
And the humans? They're crawling all over the place, dressed in cloth "bunny suits" that keep dust off the disks. Although the chipmaking is highly automated, it's hard to get it right: Engineers and R&D scientists constantly monitor every step in the process and make adjustments or experiment to increase the number of usable chips on each disk.
In most chip plants, those who aren't employees typically wear different-colored bunny suits to distinguish them. Not so in East Fishkill. All the scientists and engineers-2,000 IBMers and hundreds more who work for partners-wear the same white outfits. They work together without regard to who issues their paycheck. Sometimes an IBM worker leads a team; other times it's somebody from AMD or Freescale. "We don't work in silos," says John Pellerin, the top AMD manager at the plant. "We're a fully cross-mixed team."
Shrinking chips. Soaring costs. An industry learns to share and save.
In a business where new fabrication plants and advanced research are frightfully expensive, collaboration is nearly a necessity. Only Intel and a handful of other giants can afford to go it largely alone.
How expensive is chipmaking? A new factory costs $4 billion or more over two years. Then there's the research into designs, materials, and chipmaking processes needed for a company to remain competitive. R&D costs have been rising at an average of 12% per year while revenues for the industry are growing at just 6%. "The cost of research is kicking them," says Dan Hutcheson, CEO of vlsi Research. Collaboration saves billions. All told, Hutcheson estimates the members of IBM's chip alliances have saved $2 billion to $4 billion on research costs by throwing in together. Market research outfit In-Stat estimates the group will save an additional $7 billion over the next three years.
Innovation networks can be tough to manage.
If the benefits of collaborative innovation are easy to spot, so are the pitfalls. Think how hard it is to get people in a single corporation on the same page. Now multiply that by a factor of three or five. Who's in charge? Who owns the innovations? "The complications are always going to be about control and the danger of goals diverging," says Laura Tyson, professor of economics and business at the Haas School of Business at the University of California at Berkeley. "The other big issue is the intellectual property that you create in common. Later, if you break up, you have to be prepared to work through the divorce."
Experts say the secret to successful alliances is agreeing on common goals and setting rules of engagement from the start. Then the partners should set up procedures for day-to-day interactions, including spelling out what can be discussed by people from different companies and what's strictly off limits.
Just because a company weaves partners into its operations doesn't mean it won't have to change the way it operates. Once P&G CEO A.G. Lafley set a goal of going outside to find half of all innovations, the people he put in charge realized they would have to set up an external department to cherry-pick innovations and bring them into P&G. An in-house team of more than 200 now sizes up more than 2,500 innovations a year. "You have to set up an internal structure so you can digest all this stuff," says Larry Huston of strategy consultancy 4inno, who formerly managed P&G's external innovation programs.
In some cases, to avoid conflicts companies target fundamental research they're willing to share, even with rivals. This approach is starting to catch on in the pharmaceutical industry. Lilly and other drugmakers have teamed up to identify "biomarkers," substances that indicate the presence of a disease. With that research in hand, each can separately develop drugs to combat the diseases.
Innovation networks do sometimes fail, most often because companies' interests diverge. That's what happened earlier this year when the three partners in the Crolles2 chip research alliance went their own ways. Netherlands-based nxp Semiconductors was the first to quit, followed by Freescale and STMicroelectronics. The three companies' priorities were out of whack. Being partners "is like being roommates. If one comes in at 3 a.m. and another gets up at 7 a.m., it just doesn't work," says Andreas Cuomo, chief strategy officer for STMicro.
How IBM learned to mix cultures and make open innovation pay off.
It's telling that when Crolles2 fell apart, STMicro and Freescale quickly turned around and joined IBM's alliances. Freescale, in fact, joined three of the five. The reason: The Austin (Tex.) company has seen IBM gradually hone its skills at managing such arrangements. "This is not IBM's first rodeo," says Freescale Vice-President Gregg Bartlett. "They have a lot of experience, and there won't be any surprises."
IBM's success is hard-won. Some of its earlier collaborations didn't go smoothly. In the 1990s, the Armonk (N.Y.) behemoth formed a venture to develop memory-chip technologies with Germany's Infineon Technologies and Japan's Toshiba, but their national and corporate cultures clashed. At one point, Toshiba engineers accused IBMers of withholding information from them. The problems got so bad that the companies sent 10 people each on a three-day team-building session.
Face-to-face with their differences, the engineers mapped out better ways to work together. At IBM, people typically reached decisions by discussing problems in open meetings. Toshiba's engineers preferred to see presentations, read reports, and make decisions later. IBM's dearth of reports made the Japanese engineers suspect they were being kept in the dark. The solution: assigning people to take notes on the meetings and issue reports later.
These days IBM managers are particularly sensitive to cultural issues. Mukesh Khare, a project manager, says that typically in group discussions, Toshiba engineers will say "yes" to signal they understand a proposal-not necessarily that they agree. Later, he'll circle back to them and find out what they really think.
Mutual respect is also key to making these link-ups succeed. When IBM, AMD, Sony, and Toshiba worked together on a semiconductor breakthrough called low-k, metal gate, which makes it possible to place circuits closer together on a chip, each company brought a particular expertise. AMD, for example, was adept at devising experiments to test the group's theories. "I used to have trepidation about these relationships, but not now," says IBM Fellow Dan Edelstein. "Their work was unassailable."
Motto For the 21st Century: Network or Die.
Alain Kaloyeros isn't your typical college professor. The deeply tanned, 51-year-old physics professor at the University at Albany drives a Ferrari F430 F1 Spider with a 500 horsepower engine and a vanity plate that reads: Dr Nano. Readying for a BusinessWeek interview, he tore the tops off of 15 packets of Splenda and poured them into a 16-ounce cup of coffee. "I need sweetness in my life," Kaloyeros explained.
This high-octane prof is one of IBM's key partners in its chip research network. And he embodies two of the most important requirements of successful alliances: nerve and commitment.
In the late 1990s, Kaloyeros and IBM's John Kelly dreamed up a plan to make the Albany campus of the State University of New York a hotbed of semiconductor research. The pair relentlessly pursued their vision until they got the state and corporate funding they wanted. Now, 10 years and $4.2 billion later, Albany Nanotech boasts a staff of 1,800 university and corporate scientists and is the most advanced university chip research complex in the world.
There, Big Blue and its partners gain access to the latest chipmaking equipment and design the processes that they'll use when those machines are installed in their own manufacturing plants. That gives them a head start on other companies that can't afford such early access. And it gives members of the alliance a chance to make up ground on Intel, which, thanks to its vast resources, is typically a year ahead in advancing to each successive generation of technology.
The hookup between IBM and New York State has been something of a high-wire act. Both sides ran into funding hurdles. Now the major investments have been made and the rewards are rolling in. For those who fought the funding wars, the formula for success is simple: Share the risk and stick to the vision. "John Kelly went back to IBM numerous times and said, 'This is going to happen.' He really stuck his neck out," says former New York Governor George Pataki, who backed the project. "Fortunately, the state came through."
IBM's alliances with the likes of Albany Nanotech, AMD, and Freescale have paid off just the way its leaders hoped. Now, the company is expanding its innovation ecosystem to include suppliers of chip materials, chemical companies, and chip-design software companies. "This is a model that will not only survive but will prosper," predicts Kelly, who is now director of IBM Research.
For pioneers such as Kelly and Kaloyeros, there's no turning back. For other R&D leaders, an open-innovation strategy is still new and risky. But as more companies embrace it, the pressure will be on the holdouts to reach across organizational borders in search of ideas and greater productivity. They can delay, but they could be left far behind if they don't play.