Thursday, November 12, 2009

Money & Business

USN Current Issue

Chipping Away

In a digital world, Texas Instruments finds profit in analog

By David LaGesse
Posted 3/11/07

Scattered around some sections of Texas Instruments are meeting places commonly known as "beer check" rooms. No beer gets consumed there, at least that engineers are willing to admit. The suds instead are an incentive-engineers win a case of their choice, root beer if they must, when they pinpoint a fatal flaw in the blueprint for a new semiconductor. "A design gets put up on the wall, engineers are summoned, and they have at it," says Bill Klein, a TI engineer who has long made a living in a little-understood corner of semiconductors known as analog chips.

In a world gone digital, analog chips are a surprising champion. They are the semiconductors that handle power, light, and sound-turning those real-world phenomena into virtual reality, into the numbers where digital chips work their math magic. They're also where Texas Instruments, a venerable name in semiconductors (where the integrated circuit itself was invented nearly a half century ago), has bet much of its future. The Dallas-based company was for a quarter century the largest maker of semiconductors but found itself stumbling in the early 1990s. So it underwent a transformation, shedding all sorts of business lines to focus on two-digital signal processors and analog chips. The two products accounted for less than 20 percent of TI's sales in 1996 but now make up about 80 percent. And of the two, the future could increasingly rest on analog, says TI CEO Richard Templeton: "Despite the world going digital, the growth of analog is going to be very, very powerful."

Templeton looks the part of an engineering CEO, dressing stylishly in sport coat, jeans, and black sweater, clean-shaven with close-cropped hair. So it's all the more intriguing to know his company is betting on the likes of Klein, who offers a glimpse of silicon's wilder side, with a silver Stetson that's always with him and an ankle-length "Great Plains duster" nearby. This is not the staid world of digital semiconductors, where legions of young, buttoned-down engineers stare at computer screens as they assemble the next blockbuster microprocessor. At 65 years old, Klein with his full gray beard is a more common type in the analog world, where intensely independent, beer-betting designers often work alone in devising their circuits, where experience is golden and intuition the silver of a highly profitable high-tech business.

Turnaround. While digital can involve millions of circuits squeezed onto one sliver of silicon, analog might mean a few dozen carefully arranged to minimize interference or heat. "It's really kind of a black art," says Robert Burleson, an equity analyst at ThinkEquity Partners. "I envision guys in Led Zeppelin T-shirts and ponytails working on these chips in their basements." It's a valuable art as well, with Burleson and other analysts enthusiastic about TI's progress in analog, where the company has steadily been grabbing market share. Analog is key to a turnaround at the Dallas company, which in the 1980s had fallen behind its semiconductor competitors, notably Intel, which increasingly dominated the silicon used in personal computers.

Under former CEOs Jerry Junkins and Tom Engibous, now TI's chairman, Texas Instruments in the 1990s shed more than a dozen of what it considered old businesses, including defense products and oil-field services. The latter is where the company started in the 1930s. It also abandoned the PC market, including notebook computers, microprocessors, and memory chips. "We moved beyond the PC, because we saw the market moving beyond it," Templeton says. Still, TI went through painful times as the rest of the tech world thrived amid the 1990s Internet boom and its own fortunes sagged; it lost money while shrinking in size by more than a quarter, from sales of about $11.4 billion in 1995 to $8.5 billion three years later.

In doing so, the company remade itself to focus on what Engibous and Templeton saw as the future: wireless handsets, where it already had a major play in its digital signal processors. DSPs are integrated circuits with roots in the speech synthesizer that powered the Speak & Spell, a successful toy that TI started selling in the late 1970s-and that E.T. hacked to communicate with fellow extraterrestrials. A DSP takes digital streams, such as audio and video, and manipulates them, scrubbing them of background noise or squeezing them for quicker transmission. With money from its downsizing, TI bought numerous companies that reinforced its push into DSPs and, later, into analog. The transformation "not only took foresight; it also took nerve," says Ron Slaymaker, TI's head of investor relations.

The effort paid off by the late 1990s as cellphone sales skyrocketed, and the company dominated DSPs, with about half the market. That figure now exceeds 60 percent, but the rocket is beginning to show cracks: Some analysts worry that the cellphone market is nearing saturation and poised to slow down. "Then again, we were saying the same thing three years ago, and it hasn't happened yet," says Slaymaker. Still, there is at least a temporary dip underway, as more cellphone sales move overseas, where cheaper handsets are more popular; TIhas acknowledged that the dollar value of its sales may be soft for this quarter and maybe longer.

But Templeton argues that the company has positioned itself well for cellphones in emerging markets like India, China, and Latin America, particularly with a cheap, one-chip solution it calls the LoCosto. "Very sophisticated name, isn't it?" the CEO says with a smile. "But people aren't confused about its purpose."

More sophisticated is the product's integration of DSPs, for processing the digital sound streams, with analog chips that convert voices to and from digital. And that's at the heart of the company's push into analog, which TIsaw as a chance to leverage its dominance in DSPs. Analog and digital chips pair in all sorts of electronic devices, from cable modems to glucose monitors to a PC mouse. On many, TI can now offer to sell a manufacturer 80 or 90 percent of the silicon needed. "It gives us great ability to walk around the block," says Gregg Lowe, senior vice president over TI's analog business. He's not talking recreation but to walk potential sales around a block diagram-a graphical representation of the pieces to be assembled into an electronics product.

That means efficient sales calls, with TI's large marketing group able to offer a broad catalog of products to any electronics maker. "They do a very good job of cross-selling," says Susie Inouye of Databeans, a company that tracks semiconductor sales. Selling DSPs alone would still be a good business, but in some ways they are what microprocessors were in the '80s and '90s. The digital chips are primarily rated for speed, like the megahertz race that finally petered out for PCs. "You can only go so fast and people still say, 'Wow,'" Inouye says.

Talent roundup. Analog chips, on the other hand, are more diverse, rated on a variety of specs, and used broadly, if for nothing else than controlling the power to gadgets. "I can guarantee you that every electronic device made-if it has a battery or plugs in-has at least one analog chip," says Templeton, the CEO. So through a series of major acquisitions from 1999 through 2001, TI assembled an organization of 3,000 workers, building on an analog group that already existed. One purchase, of Tucson, Ariz.-based Burr-Brown Corp., still ranks as the largest semiconductor merger ever at a cost of $7.6 billion. That combination added 1,500 employees, including cowboy-dressing Bill Klein, and their expertise in cutting-edge design, what the industry calls high-performance analog.

Marketed in ads that feature race cars and speed boats, high-performance chips bring big profits-with operating margins at an eye-popping 40 percent or better. Customers typically worry more about performance and reliability than they do about price, says Lowe, the analog chief. "They're not price elastic but specification elastic," he says. But analog is a woolly market. Many companies produce the chips, and nobody dominates. With its acquisitions, TI sells perhaps 14 percent of analog chips, and that's only a percentage point or two more than other companies. In the prized high-performance market, key competitors include Linear Technology, Maxim Integrated, and National Semiconductor, companies that do little but analog chips. The high-performance circuits are highly specialized, making it difficult to manage at a large company like TI, which has to produce about 500 new products a year to keep pace. The ability of specialized competitors to develop high-performance chips that outdo TI's is one danger for the chip giant. Another is in the creative and independent sensibilities of the designers.

Stories abound of analog idiosyncrasies. National Semiconductor guru Bob Pease, another graybeard, is known for heaving PCs off buildings in moments of digital pique. "We don't trust computers in designing analog chips," he says. "You have to rely on your instincts." Electricity entering one side of a chip might affect other circuits with interference, or generate heat that sends temperature changes flowing across the silicon in ways that can't be precisely predicted. Analog requires a touch that's developed over time, so much so that the chips are particularly susceptible to engineers openly, or secretly, signing their works like pieces of art.

Some of the eccentricities have faded over the years, says Gary Grandbois, a former analog designer and now a market analyst at iSuppli. But engineers still take pride in their creativity. "I'm a musician," says Steve Parks, almost coming out of his chair, having already displayed ample enthusiasm for his job as head of marketing for TI's analog business. Looking a bit like John Denver, Parks was once a chip designer and, before that, an aspiring Elton John or Billy Joel-a keyboardist. He keeps up with his music, which is typical of analog engineers, he says. Analog means waves, whose ever changing shades of gray differ from the certainty of digital, with its on-off, ones and zeroes. "Musicians know how to do improvisation," he says. "In analog, knowing how to solve a problem takes that kind of DNA."

Intuition and hunch are mentioned often. "Ideas come together much more often on a whiteboard, or even the back of an envelope, than in digital," says Brett Forejt, an analog designer at TI. Experience counts, giving older designers more sway and job security than in other semiconductor jobs and raising the risk for a company like TI. Analog designers are known to leave when they feel beset by corporate folderol. "When they move, they tend to move in droves," says Inouye at Databeans. "That keeps the companies scared." A group of designers, in fact, left National Semiconductor some years ago to found Linear Technology. Still, at $1 billion, Linear is a much smaller company than TI, which had sales of $14 billion last year. Texas Instruments, says Inouye, is the biggest company with a major analog play.

Lowe, TI's analog chief, was charged with keeping the expanded analog division from coming apart. "I went home many nights wondering if it would work," he says. Six years later, it seems to have held together, with 90 percent retention of the new employees.

TI's manufacturing capability seduced some of them, as it often was more advanced because of its work in digital semiconductors. Managers moved analog chips into former digital plants, and designers found they could push their concepts further. Another benefit for Texas Instruments: Fabrication gear for analog chips can be used for 10 or 15 years, versus perhaps two for digital semiconductors. "The analog chip business is capital efficient," says Slaymaker, the investor relations chief.

But making the mergers work, and holding together the analog group, depended on a hands-off approach, he says: "We did not come in as an acquirer. We left management teams in place and told them we wanted to learn what they do." So Burr-Brown, for one, remained largely intact, including its creative processes. Designers stayed in Tucson, cowboy hats were embraced, and the beer checks continued.

This story appears in the March 19, 2007 print edition of U.S. News & World Report.

Use of this Web site constitutes acceptance of our Terms and Conditions of Use and Privacy Policy.