Winners take all

Texas Instruments, the state and the environment all got what they needed from a new 300-millimeter wafer fab

By Bill Roberts -- EDN, October 1, 2006

Time will tell if Texas Instruments' new 300-millimeter wafer fab lives up to its advanced billing, but at the moment, it's hard to imagine that there's ever been a bigger win-win-win capital investment project in electronics. Called RFAB, the 1.1-million-square-foot plant on a 92-acre site in Richardson, eight miles north of TI's Dallas headquarters, is a study in enlightened self-interest on the part of TI, the state university and state and local governments.

"We couldn't have done this project without community support and without state support for the University of Texas at Dallas (UTD)," says Phil Ritter, a TI senior vice president. "As a company, you get involved in the community, because it is the right thing to do and to achieve an important business outcome. Those came together in this project."

No single factor decided the matter, but local tax abatements, the state's commitment to upgrade the local university and TI's desire to keep the fab close to R&D in Dallas led to TI's decision to build the plant in Texas.

Instead of pressuring the state to throw money at it to keep the plant in Dallas, TI urged the state to spend $300 million to transform UTD's engineering school into one of the nation's top programs. TI will hire many graduates and benefit from advanced research done there, but so will others. Dallas reaped an infusion of construction jobs and, thanks to the UTD upgrade, the region has a chance to create a semiconductor ecosystem like it has never had. When fully ramped up in about 18 months, RFAB will employ a thousand people, but an economic impact study estimated 75,000 new technical jobs in the region from the ripple effects.

"If RFAB had not happened, the semiconductor industry in the Dallas area would be in a death spiral," says Phil Wilson, the governor's deputy chief of staff.

There's more. Because TI executives gave engineers time to innovate, they built a plant with 30 percent more capacity than its predecessor for the same price. TI also expects its sustainable—green—design to cut yearly operating costs as much as $4 million.

The plant is a benchmark for the industry. At a time of growing concern over U.S. competitiveness because manufacturing is moving offshore, the project illustrates how business and government can work together to keep and create jobs, how sustainable design complements cost efficiency and how innovative management can foster solutions for the 21st century.

"RFAB is a microcosm of what state and local governments need to do to address the country's competitiveness challenge." —Phil Ritter, Texas Instruments

TI, the third-largest chip company, outsources 20 percent of its manufacturing, but its strategy differs by technology. For digital-signal processors and other advanced logic needing the latest geometries, TI splits the work 50-50 between its own fabs and its foundry partners. Internal production is done at one 300-mm fab (DMOS6) and two 200-mm fabs in Dallas. Sharing tools, test equipment and R&D is advantageous, says Kevin Ritchie, senior vice president of technology and manufacturing. "There is a tremendous amount of synergy in having these colocated fabs."

TI makes 90 percent of its mixed-signal analog and all high-performance analog products in-house at its older fabs worldwide. By pushing mature technologies to older fabs, it gets higher capacity utilization and wrings maximum return from capital investments.

"TI's strategy is working well," says Dean Freeman, a Gartner analyst. "The company has not had to ramp up fabs as fast as anyone else because it has managed its capital investments well. This strategy allows it to keep its own fabs fully loaded."

RFAB continues the strategy. "We're looking at this new fab for 65 nanometers and will definitely need it for 45 nanometers," says Ritchie. "At the latest, 45 nanometers will qualify at our DMOS6 300-millimeter fab in 2008, and we'll bring it to RFAB after that." The company already runs 65 nm at DMOS6 and at one of the 200-mm fabs in Dallas.

CEO Rich Templeton says customers are the big winners. TI strives to bring leading-edge chips to market faster than its rivals, giving customers advantages over competitors. TI cuts development time when process R&D and advanced fabs are near each other. By keeping RFAB in Dallas, where it has extensive R&D resources and other fabs, TI will remain ahead of rivals for years to come, he believes.

"As we faced the choice of where to put the next advanced CMOS facility, the thing we weighted as being most important was time to market," Templeton says. "It helped that we had a lot of experience in doing R&D halfway around the world and knew how difficult that can be."

Balancing cost and talent

In 2003 TI executives began planning seriously to add capacity. The company hoped to keep a new fab in Dallas, but with costs rising, it had to consider other sites. The fab would cost $3 billion to construct and equip, 50 percent more than DMOS6 in 1996, says Ritter. Equipment makes up the largest portion of cost, so the two variables in the decision were other costs—construction and operational—and access to talent, he says.

" In challenging the team, we gave it time to do the design and concept work." —Kevin Ritchie, Texas Instruments

China is least expensive, but the U.S. does not issue licenses to export advanced technology there. Singapore, however, is fair game; labor costs would be half, and it offered a 10-year tax break worth a billion dollars. "Singapore was a serious contender, but not for long," says Bob Goforth, a partner at Leek & Goforth, a consulting firm that assisted in site evaluation. "TI couldn't see doing R&D that far away."

In the states, Goforth narrowed the list to New York, Virginia and Texas. New York's financial package was the richest, he says, but the Richardson property, which TI had acquired years earlier, remained the favorite. Still, TI executives had concerns.

The state was facing a school finance crisis when the legislature convened in 2003. Every aspect of the tax structure was to be scrutinized, including tax abatements, tax breaks on capital equipment and an R&D tax credit. TI needed—and got—guarantees from the governor and legislators that existing capital investment tax incentives would not be revoked.

With tax concerns settled, talent became the key issue. Albany, N.Y., and Richmond, Va., had improved their engineering schools. For years TI executives had been disappointed that Dallas was one of the only large metro areas without a top engineering school. As an incentive to build the fab in north Texas, TI wanted an engineering school that would do basic research and train world-class graduate-level engineers—future TI employees.

Says Ritter, "There is a symbiotic relationship between academic and commercial R&D. Ours tends to be short-term-focused. The longer term is important, and we rely on universities. The more of that we have in our backyard, the more competitive we will be."

At the time, UTD was looking for a new engineering dean. The search panel included Bob Helms, a TI researcher, former Stanford professor and past Sematech president. It soon became clear that Helms was the best man to be dean. He drafted a plan and a $300 million budget to turn UTD into a top school and agreed to be dean if the state adopted his plan. The state agreed, if TI would build RFAB in Dallas—and the rest is history.

Governor Rick Perry earmarked $50 million for UTD from a $300 million fund for clinching economic development deals. He recently also gave $15 million from the fund to Samsung to build a 300-mm fab in Austin. TI doesn't get a cent of the $50 million, which all goes to UTD in installments as the university and TI achieve milestones (see "Academia focuses on industrial needs," page 34).

"If there had not been a lot of trust among all parties, this would not have happened," says Wilson. "You must collaborate in a transparent win-win environment."

Says Ritter, "This project is a microcosm of what state and local governments need to do to address the country's competitiveness challenge. The federal government is not off the hook; it still needs to fund basic research."

The deal also includes new collaboration among TI, the university, a local community college and the public school district. RFAB was not the first capital investment deal to include education, but it was the first time an entire incentive package was offered to create a world-class engineering school in a company's backyard.

Sustainability meets cost efficiency

Likewise, RFAB is not the first fab to conserve energy. Chip companies have been adopting best practices identified by the International Sematech Manufacturing Initiative (ISMI). What's different is the sheer number of methods at RFAB, the fact that TI included them as design requirements and how it melded sustainable design and cost efficiencies.

"RFAB will stand as a milestone in the design, construction and operation of fabs," says James Beasley, on loan from Intel and manager of resource conservation projects at ISMI. "I also commend TI for its openness about what it's done. Everyone benefits."

The priority was to lower costs, not to achieve sustainable design. In 2001 Ritchie told his staff, "If you can design and build it with 30 percent more production capacity than DMOS6 for the same construction cost, then TI will likely keep the fab in north Texas."

"When you first heard that number, you didn't think it could be done," says Paul Westbrook, TI's sustainable development manager. "If it could be, someone would already have done it."

Ritchie knew that the only way to achieve the goal was to design from scratch. Most chip companies, including TI, find a design that works and replicate it to save time and money, to the detriment of innovation. Now TI needed a new concept. Against that backdrop, four teams went to work on cost, technology, reliability and schedule. With the industry going through one of its worst down cycles, they had a couple of years before any decisions would be made to actually build.

"Starting with a clean sheet of paper is very unusual," says Freeman. "DMOS5 and DMOS6 were similar. Now this fab is completely new and different. Someone at TI got very enlightened."

Westbrook, whose home uses passive energy, had been a sustainability proponent for years. With TI designing from scratch, he suggested that sustainability might help cut operating costs. Ritchie agreed to a fifth team, led by Westbrook, to look into it.

Ritchie admits he was dubious: "It had to have a payback. We were not going to do sustainability for the sake of sustainability."

Even Westbrook wasn't sure, but he knew a fab's lifetime operating costs can be higher than construction costs. "I thought sustainability would probably be slashed because it cost more. But cost reduction and sustainability ended up working together beautifully."

In 2002 Westbrook's team benchmarked methods and collected energy-use data for fabs and offices. It produced several white papers, but none fully convinced Ritchie. He took a tour of Westbrook's home in the fall of 2003, just before TI and the state announced the deal. Ritchie recalls his epiphany: "Paul showed me all the sustainable pieces. They were nice. Then he showed me his utility bill. It was $130 a month; mine was $700. I thought then that maybe there was something to this."

When a company constructs cookie-cutter fabs, sustainability can be an added expense and hard to justify. Because RFAB was designed from scratch, however, sustainability could be a requirement from the start. Sustainability did not cut the $300 million in construction costs, although the project did hit the desired level—30 percent more capacity than DMOS6 for the same price. But TI estimates that the sustainability measures will save $1 million in the first year and up to $4 million annually when the fab is fully operational. That money goes directly to the bottom line.

The engineers often found that a cost saving in one place made sustainability possible, and vice versa. For example, previous TI fabs needed three stories to hold all the support tools. At 300 mm, the industry has downsized the footprint for tools. That fact and moving some electrical units outside allowed TI to eliminate an entire floor, a savings of $40 million and the largest single cost reduction. Removing a floor would also reduce heating and cooling costs. The outdoor electrical units cost a bit more but also reduced the heat produced inside. The TI team worked closely with a team of consultants from the Rocky Mountain Institute, a non-profit organization working on a more efficient use of natural resources (see "Chiller units lead to energy savings," page 36).

Says Templeton, "As an engineering culture, we believe that waste is cost. Many things labeled 'green' are about waste, especially electricity and water. Whatever sustainable design can do for the planet, it also makes good economic sense. In the end, you get both sets of benefits."

Innovative management

Underlying this project is innovative management that gave workers a challenge and then got out of the way to let them achieve it. "I didn't know how we would accomplish the cost reduction. I just knew that we couldn't carry forward previous fabs," says Ritchie. "In challenging the team, we gave it time to do the design and concept work. If we had rushed the design, we would have gotten another DMOS6."

There were enlightened management decisions like that throughout the project. Project manager Terry Dalton assigned only engineers who had some people skills. "There was a list in the back of my mind of people I would never have picked," Dalton says. "People make the big difference, being open-minded and being aggressive."

That was crucial, says Westbrook. "Engineers tend to be optimizers, but no one had said let's optimize the whole problem," Westbrook says. "That required understanding each other's viewpoints. The challenge was getting people to talk to each other. So everyone selected for this project had some people skills."

At Ritter's level, the deal with the state and the university was the result of years of relationship building. Ritter is a 15-year veteran of cultivating relationships with the state and Dallas, having worked on several other capital investment projects.

Freeman, who worked at TI in the mid-1980s and has watched it as an analyst for two decades, sees an evolution in the company's management style. The TI he worked for would never have produced an RFAB. "Each generation of TI management seems to have become more forward-thinking," he says.

To what does Templeton attribute this?

"Our products, technologies and customers have changed dramatically through the years, but our belief in innovation has lasted."

Bill Roberts is a contributing writer at Electronic Business.