Advanced Micro Devices’s Lisa Su is an ‘engineer’s engineer’ who just muscled in on the battle for the newest technology in the world.

As a child, Lisa Su opened up her brother's remote-controlled car when it broke, and, as she tells it, identified a loose wire connection and taught herself how to repair the toy.

In 2021, she became the first woman to be awarded the Robert N. Noyce medal, considered the Nobel Prize of microelectronics. This week, she muscled in on the battle over the newest and most advanced technology in the world that has been dominated by boldfaced names like Sam Altman and Jensen Huang.

Su pulled off a megadeal with OpenAI, unveiled Monday, that positions her company, Advanced Micro Devices, to take on archrival Nvidia. Now the market is watching for her to strike additional deals and sign up more big artificial-intelligence firms as customers for AMD's chips. The deal caps Su's decadelong turnaround effort.

As a CEO in the old Silicon Valley mold -- an engineer who intimately knows the technology she's selling -- Su is deftly navigating challenges that have tripped up other tech-industry leaders. While rivals like Intel got bogged down in costly manufacturing investments, Su focused squarely on what she believes AMD does best: designing ever more powerful semiconductors. She has transformed AMD into the most serious competitor to Huang and Nvidia, the 800-pound gorilla of the AI chips industry.

Since taking the helm in 2014, she's pulled AMD back from the verge of bankruptcy and built the company's market value from less than $3 billion to an astounding $348 billion.

Under the terms of this week's agreement, Altman's OpenAI, the maker of ChatGPT, is expected to buy 6 gigawatts worth of advanced AI chips, handing AMD tens of billions of dollars in revenue over the next few years. The company's share price rose more than 33% this past week on the news.

"If you ask, what do we do when we wake up in the morning, it's about, how do we push the envelope in high-performance computing?" Su recently told the venture capitalist Reid Hoffman in a podcast interview. "And it turns out that today, that high-performance computing is all about AI."

Born in Taiwan to a mathematician father and accountant mother, Su, 55 years old, immigrated to the U.S. in the early 1970s, when she was 3 years old. Her family settled in Queens, N.Y., and she attended the prestigious Bronx High School of Science. Su is a distant cousin of Huang, the co-founder and CEO of Nvidia, but didn't meet him until well into her career.

She later obtained bachelors, masters and doctoral degrees in electrical engineering from the Massachusetts Institute of Technology, where she grew interested in computer chips while working in a semiconductor lab as an undergraduate.

Early in her career, she worked at IBM, helping design some of the earliest copper-wire interconnections for semiconductors and learning how to bring new technology to market as a product manager. She joined AMD in 2012 as a senior vice president and quickly rose to chief operating officer.

Su has described herself as an "engineer's engineer" who is fascinated by device physics, a discipline that studies how machines can interact with tiny particles like electrons and protons. Her understanding of the nuts and bolts of microelectronics has primed her to lead AMD at a time when building AI supercomputers requires finding increasingly inventive ways to cram billions of transistors that are a fraction of the width of a human hair onto postage-stamp sized slices of silicon.

"At the end of the day, the beauty of engineering is, it actually matters what you do," Su told Hoffman in her interview.

Although AMD is based in Santa Clara, Calif., Su lives with her husband in Austin, Texas, and commutes to Silicon Valley several times a month.

Just before last Christmas, SemiAnalysis, an influential tech research shop, published a long, technically complex article comparing AMD's newest AI chips unfavorably to Nvidia's. While AMD's hardware is sophisticated enough to compete with its bigger rival, the article said, its software platform was full of bugs and impossible to use in large-scale AI training.

The day after it published, Su called up the author, SemiAnalysis founder Dylan Patel, and spent an hour and a half going over his arguments. She acknowledged problems in AMD's products and listened to Patel's detailed recommendations for how to fix them. Afterward, she thanked Patel for his help in a social-media post.

"Feedback is a gift, even when it's critical," she wrote.

AMD was founded in 1969 by Jerry Sanders and other veterans of Fairchild Semiconductor, a pioneering developer of what were known as integrated circuits, the forerunners of modern chips. For the next two decades, AMD was consistently playing catch-up to Intel, another spawn of Fairchild that created the microprocessors that were used in most personal computers starting in the 1980s.

When Su took over as CEO, AMD had "lost the plot" when it came to data-center chips, and its market share was near zero percent, said Forrest Norrod, AMD's head of data-center products and her first hire as CEO. Su developed a systematic approach and today AMD has a market share of about 41% in data-center CPUs, the central computer brains that power servers used for cloud computing and AI functions, he said.

"AMD before Lisa was trying to chase the bulk of the market, the mainstream parts rather than high-performance parts," Norrod said. "She understood that our customers care the most about the high-performance parts, and if you're focused on the bulk of the market, then your competitors can outmaneuver you."

"When you're talking to her, she'll test you on what you're thinking both on the business side and the technology side," he said.

She quickly set teams to work designing new graphics cards for videogames -- AMD's chips power PlayStation and Xbox consoles, among other gaming systems -- and more powerful data-center and PC chips, steadily eating away at Intel's market share. Intel meanwhile fell behind.

AMD's biggest foray into artificial intelligence began in late 2023 with its launch of the MI300 series of graphics processing units, or GPUs, the chips required for the advanced computing workloads behind AI.

Nvidia controls more than 75% of the AI data-center chip market, most analysts estimate, and has a major head start over rivals, in part because of its proprietary CUDA software system, which provides thousands of developers with a streamlined programming language that they can use to create applications using only Nvidia's chips.

Huang, the Nvidia CEO, has praised Su for her AMD turnaround and called her "terrific." This past week, however, he questioned the wisdom of her new deal with OpenAI, under which OpenAI stands to receive 160 million shares of AMD stock in phases if certain milestones are met.

"Considering they were so excited about their next-generation product, I'm surprised that they would give away 10% of the company before they even built it," Huang said of AMD in an interview on CNBC. "Anyhow, it's clever, I guess."