After eight and a half years at MIT, Lisa Su had earned three degrees in electrical engineering and was eager to get her life started. But before she left Cambridge for her first job at Texas Instruments, her advisor, Dimitri Antoniadis, gave her a bit of career advice. “Stay technical as long as you can,” he told her. “Once you leave it, you’re never going to be able to operate at that same level again.” Su listened, perhaps even nodded, but then went on to prove him wrong.
“People feel like you have to make a choice” between being a serious researcher and a businessperson, says Su. She disagrees. And her meteoric rise at IBM, where she’s now vice president of the semiconductor research and development center, is proof that it’s possible, and in her case even advantageous, to choose not one or the other but both. “I find the ability to go back and forth very useful,” she says. “It is very, very rewarding to be a deep expert in one area. I just feel like that’s one path. I enjoy more being a moderate expert in lots of different areas.”
Su is as comfortable dealing with CEOs as she is having technological discussions with university professors and IBM research fellows. (“I have to convince both the CEOs and the deep-tech folks that we know what we’re talking about,” she says.) And by maintaining her technology expertise even as she has taken on business leadership responsibilities, she has been able to serve as a translator between two very different worlds.
“Sometimes deep technologists find the business and the strategy stuff boring,” she says. “But I don’t.”
In the fall of 1986, Su arrived at MIT from the Bronx High School of Science intending to major in electrical engineering or computer science; after taking “weed-out classes” 6.001 and 6.002, she chose electrical engineering because she found it harder. As a freshman, she landed an Undergraduate Research Opportunities Program (UROP) assignment in Hank Smith’s semiconductor lab in Building 39, making two-inch wafers for his x-ray lithography research. It was pure grunt work, she says, “but as an undergraduate, I didn’t know; it was great.” That UROP experience and summer jobs at Analog Devices got her interested in a technical career in semiconductors. “At the time, I had so many colleagues who were going to Wall Street or taking their technical background and applying it in other fields, it was a pretty big decision to stay more technical,” she says.
As a doctoral candidate, Su was one of the first researchers to look into silicon-on-insulator (SOI) technology, a then unproven technique for increasing transistors’ efficiency by building them atop layers of an insulating material. It was “pretty exciting stuff,” she says. “The application of SOI right now is very important in microprocessors. At the time, it wasn’t so clear what the right application was.”
Today, SOI is used either to boost microchips’ performance by up to 30 percent or to significantly reduce their power requirements. Although Su’s doctoral research turned out to be far reaching, she insists what’s important about a PhD isn’t the project you work on. “It’s not supposed to be job training,” she says. “It’s the confidence that you build. When I graduated from MIT, I felt like I was one of the world’s experts on SOI devices. And that was a great feeling.”
PhD in hand, Su spent a year at Texas Instruments before joining IBM in 1995. At Big Blue, she got assigned to a project looking into how to replace semiconductors’ traditional aluminum interconnects with speedier copper ones, without having copper impurities contaminate the chips during production. “My specialty was not in copper,” she says, “but I migrated to where the problems were.” Su worked with IBM design teams to hammer out the details of the device design. Once she thought the technology was mature, she was ready to move on and told her boss she wanted a new assignment. “I remember very clearly, he told me, ‘Nope, you’re not done until we’re shipping products,’” she recalls.
“It’s very easy to just stop before the end, because you think that all of the innovation is done,” she says. “You’re not going to write new patents in those last few months, but you’re going to learn an incredible amount of practical knowledge.” Su says the last six months before a product ships are the toughest, because that’s when you recognize – and have to address – all of the second-, third-, and fourth-order problems. “Those are the things you can’t learn in books,” she insists. The work paid off when IBM introduced copper chips that are 10 to 20 percent faster than conventional chips made with aluminum.
Once the copper chips did ship, Su got tapped to serve as technical assistant for Lou Gerstner, IBM’s chairman and CEO. “I got lucky,” says Su. She had been at IBM for only five years, but Gerstner wanted a different kind of technical sidekick, someone newer to the business and thus closer to new technologies. In the process of showing him what she calls “a few of the latest technology tricks,” Su got to see firsthand how he approached leading a large organization – and to find out what else a CEO thinks about. Gerstner thought a lot about the competition, she says.
“Lou was very interested in the technology itself; he wanted to understand it,” says Su. “So part of my job was to translate the deep technology into something that could be understood at a business level.”
The expected path for Su, upon completing her one-year assignment with Gerstner, would’ve been to go back and run a larger research organization within IBM. But she didn’t want to follow the usual career path; she wanted to learn more about the business and instead took on the role of director of emerging products. (“I was basically director of myself – there was no one else in the group,” she says.) In looking for ways to apply technology beyond PCs and servers, IBM had zeroed in on the game-machine industry, which was still dominated by 300-megahertz devices. Su soon found herself representing IBM in a collaboration with Sony and Toshiba to create next-generation chips for gaming and other applications that would last for the next 10 years. Ken Kutaragi, CEO of Sony Computer Entertainment, charged the collaborators with improving the performance of game machine processors by a factor of 1,000. He wanted power, performance, and the right price. “To tell you the truth, we came back with some evolutionary solutions, and he basically said, ‘Nah, not interested,’” Su recalls. “And we’d have to rip that up and go back to the drawing board. It took us a couple of tries.”
Eventually, the team came up with the idea of a nine-processor chip. One processor functions as the traffic cop and takes care of everyday things. The extra eight processors are optimized for handling multimedia content or doing many things in parallel. In addition to powering the vastly improved graphics of the highly anticipated Sony PlayStation 3, the Cell chip, which IBM, Sony, and Toshiba announced in 2004, will be used for things like high-speed medical imaging that require real-time visualization of vast amounts of complex data.
Su believes there’s room for more innovation in silicon. “Everyone is predicting the end of Moore’s Law. I think we still have a pretty long ways to go,” she says. “Today we’re pushing the device dimensions in a two-dimensional way smaller and faster. Yet 10 years from now, we may be doing three-dimensional integration,” stacking circuits to cram even more on each chip.
In her effort to help IBM keep pace with Moore’s Law, Su employs “a healthy bit of patience and impatience,” as she puts it. She’s very patient about learning and is willing to invest time to understand a business or technology situation in depth. But she’s impatient when it comes to getting people moving and making sure no one reverts to a silo mentality. “If each team optimizes in its own box, you’ll get one answer. But if each team can open up its box and show each other where their pinch points are, you can come up with an answer that’s much better,” she says. “It’s not that people don’t want to do that; the problem is in the translation. People don’t speak each other’s language.”
Su is an able translator because, she says, she’s always been a person of many interests. “Although you’re coming to a technology school, that doesn’t mean that’s all you’re interested in,” she says. Su thinks that a lot of MIT people are 51 percent interested in technology and 49 percent interested in other things. Concludes Su, “All I did is, I didn’t let that 49 percent drop.”
This new data poisoning tool lets artists fight back against generative AI
The tool, called Nightshade, messes up training data in ways that could cause serious damage to image-generating AI models.
Rogue superintelligence and merging with machines: Inside the mind of OpenAI’s chief scientist
An exclusive conversation with Ilya Sutskever on his fears for the future of AI and why they’ve made him change the focus of his life’s work.
Data analytics reveal real business value
Sophisticated analytics tools mine insights from data, optimizing operational processes across the enterprise.
Driving companywide efficiencies with AI
Advanced AI and ML capabilities revolutionize how administrative and operations tasks are done.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.