The Three-Dollar Company
On the day in October 2014 when Lisa Su officially became chief executive of Advanced Micro Devices, the company's stock closed at roughly $3 a share. This is the kind of number that, in semiconductor circles, functions less as a price than as a prognosis. AMD had just cut a quarter of its workforce. It had sold its Austin headquarters. Its share of the data-center chip market had fallen so low that internal executives rounded it down to zero — not as metaphor but as accounting shorthand. The company carried billions of dollars in debt, its product roadmap was barren of anything that could credibly challenge Intel in computing or Nvidia in graphics, and the industry consensus, voiced with the blunt tenderness reserved for terminal patients, was not whether AMD would die but when. Mentors — people who genuinely cared about Su's career — told her not to take the job. "I actually had mentors in my career saying, you know, I don't think that that's a good move," she recalled at Stanford a decade later, the kind of understatement that sounds wry only in retrospect.
She took it anyway. Not despite the wreckage, but because of it. "When you grow up as a tech person — and I spent my career in semiconductors — there aren't that many large US semiconductor companies," she told CNN. "So I was really excited to become CEO." She liked, she said, that she could walk around a Best Buy and pick up a laptop, knowing that the chip her company built was powering it. The romance of the thing was inseparable from the engineering of it, the way the romance of a building is inseparable from its foundation. The building was crumbling. She wanted to rebuild it.
What followed over the next decade is, by the cold metrics of shareholder value, one of the most extraordinary corporate turnarounds in American business history — a roughly 85-fold increase in market capitalization, from $2 billion to over $200 billion, a stock price that would climb from $3 to above $160. AMD would surpass its eternal rival Intel in market value for the first time. It would power the world's two fastest supercomputers. Harvard Business School would codify Su's stewardship as a case study.
Time would name her CEO of the Year for 2024. Chris Miller, the historian who wrote
Chip War, called it "one of the great turnaround stories of modern American business history."
But the numbers, impressive as they are, obscure something more interesting about how the turnaround actually happened — which is to say, slowly, methodically, with the compounding patience of someone who understands that in semiconductors, the decisions you make today don't show up in products for three to five years, and that this lag is not a liability but a competitive weapon if you are the kind of person who can think clearly across long time horizons while everyone around you is panicking about the quarter.
By the Numbers
The AMD Transformation
~$3AMD stock price when Su became CEO (October 2014)
$205B+Market capitalization by late 2024
~85xIncrease in market cap during Su's tenure
<1% → 40%+Data-center CPU market share growth
$5.5B → $25B+Annual revenue growth (2014 to 2024 projection)
$1.3BSu's estimated net worth (Forbes, April 2024)
10+Years as CEO, the longest-tenured in AMD history
Multiplication Tables at the Dining Room Table
Lisa Tzwu-Fang Su was born in November 1969 in Tainan, Taiwan, a coastal city in the island's south that has produced, among other things, an improbable concentration of semiconductor talent. Her family immigrated to the United States when she was three — or two, depending on which interview you consult; Su herself sometimes varies the number — so her father, a mathematician, could attend graduate school in New York. Her mother was an accountant who would eventually start her own import-export business in her mid-forties, linking Taiwanese manufacturers to American companies. The household values were legible, non-negotiable, and Confucian in their directness: education, achievement, the unsentimental expectation that their children would reach the highest possible plane of intellectual accomplishment. "There were expectations that you should get all A's, that you should go to the best schools and that you should get a Ph.D.," Su recalled. "They weren't questions like, 'Do you want to?' They said, 'You should do those things.'"
Her father quizzed her with multiplication tables at the dining room table when she was five. The image is almost too neat — the future semiconductor CEO, drilled on arithmetic before she could read — but Su returns to it frequently, less as origin myth than as the earliest expression of something she considers foundational: the conviction that rigorous, structured thinking is not innate but practiced, not a gift but a discipline. "That's how I first got into math," she told Forbes, with the economy of someone who does not believe in ornamenting facts.
The family landed in Queens, New York. Su attended the Bronx High School of Science — the public magnet school that has produced more Nobel laureates than most countries — and demonstrated the particular brand of restless curiosity that manifests, in certain children, as an irresistible compulsion to disassemble things. She took apart her brother's remote-controlled car when she was ten, not out of mischief but out of need. She had to see the motor. She had to understand what was underneath.
Her mother, in a story Su has told with dry fondness, gave her three acceptable career options: doctor, concert pianist, or engineer. Su auditioned for Juilliard. She did not get in. "I wasn't good enough to do that," she has said, "so I became an engineer." The line gets a laugh every time she deploys it, which is often, but it conceals something true about her temperament: a ruthless, unsentimental capacity for self-assessment, the ability to measure herself against a standard and, finding herself short, pivot without grievance to the thing she could actually be best at. This is a rarer quality than ambition. Ambition drives you forward; self-assessment tells you where.
The Hardest Major
At MIT, Su chose electrical engineering over computer science. The reason, as she has told it with disarming frankness: "Electrical engineering, particularly at MIT, was the hardest major, so I said, 'You know, how about we try that and see how it goes.'" Computer science meant writing software programs. Electrical engineering meant building things. Physical things that had to actually work. She wanted to build.
She earned her bachelor's degree in 1990, her master's in 1991, and her doctorate in 1994 — three degrees in electrical engineering from the same institution in four years, a velocity that speaks less to genius (though no one doubts her intelligence) than to the specific kind of relentlessness that Su would later bring to AMD's product cycles. Her PhD research focused on silicon-on-insulator (SOI) technology, a foundational innovation in chip design that would later revolutionize processor performance and energy efficiency. It was deep, esoteric work — the kind of research that operates several layers of abstraction below anything a normal person would ever encounter — and it taught her something she considers more important than any specific finding.
"My PhD taught me how to think about problems," she told founders at Pear VC in 2025. "I started not knowing anything about silicon on insulator devices, and by the end of the four years I was an expert in the field. That gives you so much confidence in terms of how to solve problems." The confidence she describes is not bravado. It is the specific confidence of someone who has proven, to herself, that she can master any domain if given enough time and enough data. It is a confidence rooted not in what she knows but in how she learns.
During her PhD studies, she created instructional guides for using the Institute's most advanced fabrication equipment — manuals so meticulous that they were used by generations of subsequent student researchers. "Long before she led the spectacular turnaround of AMD," MIT President Sally Kornbluth would note decades later, "Lisa Su was an MIT student who inspired and mentored her classmates." The detail is minor but revealing: even as a graduate student, Su's instinct was not merely to solve problems but to make the process of solving them legible and repeatable for others. It is an engineering instinct applied to institutions. It is also, recognizably, the instinct of a future CEO.
Thirteen Years in the Machine
After MIT, Su spent a year at Texas Instruments — a brief stint, almost a palate cleanser — before joining IBM in 1995, where she would remain for thirteen years and undergo what amounted to a comprehensive education in how large technology companies actually function. IBM in the mid-1990s was itself in the midst of one of the most famous corporate turnarounds in history, led by Lou Gerstner, the former McKinsey and RJR Nabisco executive who had been brought in to rescue the company from near-collapse. Gerstner — a non-technologist running one of the world's great technology companies — was reshaping IBM from a hardware manufacturer into a services and solutions business, and his leadership would become a template that Su internalized deeply.
Su's early work at IBM was pure engineering: she played a pivotal role in the development of copper interconnect technology, which replaced the aluminum interconnects that had been standard in semiconductor manufacturing for decades. The switch to copper enabled faster, more efficient chip communication, and it was a breakthrough significant enough to reshape industry standards. But what mattered more for Su's trajectory was what happened next. Five years out of graduate school, she was plucked from the labs and made technical assistant to CEO Gerstner himself — a role that, in IBM's corporate hierarchy, was roughly equivalent to being assigned as an aide-de-camp to a general.
"I grew up at IBM and it shaped me as a leader," Su said at a Semiconductor Industry Association event. "Five years out of grad school, I didn't have any idea what it meant to be a CEO. It was fun to teach Lou about technology. He understood how important technology was. He showed me how important it is to connect the dots. As cool as technology is, how does it affect customers?" The Gerstner apprenticeship taught her something her PhD could not: that technology is never the product. The product is the customer problem that the technology solves. The distinction seems obvious until you realize how many brilliant engineers never learn it.
She rose to vice president of IBM's Semiconductor Research and Development Center, overseeing the strategic direction of the company's silicon technologies and its joint development alliances. She founded IBM's Emerging Products division, which worked on biochips, energy-efficient semiconductors, and other frontier technologies. And she led the team that created the Cell microprocessor — the exotic, idiosyncratic chip that powered Sony's PlayStation 3. The Cell project was a collaboration between IBM, Sony, and Toshiba, and it was the kind of wildly ambitious, multi-company engineering effort that teaches you everything about coordinating disparate teams with conflicting incentives toward a shared technical goal. It was, in other words, perfect preparation for running AMD.
"I was really lucky early in my career," Su told Time. "Every two years, I did a different thing." The modesty is characteristic but misleading. Luck plays a role, certainly, but the pattern — research, development, executive apprenticeship, product leadership, cross-company collaboration — was not random. Each role expanded her range. Each forced her into domains she hadn't mastered. By the time she left IBM in 2007, she was that rare creature in the semiconductor industry: someone who could move fluently between the physics of transistors and the politics of boardrooms, between the lab and the P&L.
The Distant Cousin
There is a family connection that hovers over Su's story like a minor-key motif, never quite resolving into a full melody.
Jensen Huang, the founder and CEO of Nvidia — the company that would become AMD's most formidable rival in the GPU and AI chip markets, the company whose market capitalization would swell to nearly $3 trillion by 2024 — is Su's cousin. Distant cousin, to be precise, through their maternal grandfathers. The degree of the relationship is a subject on which both Su and Huang are conspicuously reticent; press handlers warn journalists not to probe the connection too aggressively.
Huang was born in Tainan in 1963, six years before Su, and his family also immigrated to the United States when he was young — a parallel trajectory that has become, for the business press, an irresistible narrative about a single Taiwanese family producing the leaders of two of the three most important American semiconductor companies. (The third, Intel, has no comparable family lore.) The parallel is real but its significance is easily overstated. Su and Huang built their companies in fundamentally different ways — Huang as a founder who had been running Nvidia since 1993, Su as a professional manager who inherited a burning house — and the competitive dynamic between AMD and Nvidia is shaped far more by product architectures and software ecosystems than by Lunar New Year dinners.
Still, the family connection matters as atmosphere. It matters because it situates both leaders within a specific immigration story — Taiwanese families who arrived in America in the 1960s and 1970s, pushed their children ferociously toward STEM education, and produced a generation of engineers who would come to dominate the industry that, more than any other, defines American technological supremacy. The irony is rich: the chips that power America's national security infrastructure, that train its AI models, that run its supercomputers — the chips that are now the subject of export controls designed to maintain American advantage over China — are designed by the children and grandchildren of Taiwanese immigrants.
The Burning House
Su left IBM in 2007 for Freescale Semiconductor, the Motorola spinoff that made embedded processors and networking chips. She served first as CTO, then as senior vice president and general manager of the networking and multimedia division. It was a smaller stage, a tighter operation, and it gave her something IBM could not: direct profit-and-loss responsibility for a product business. At Freescale, she wasn't advising the CEO or running R&D. She was running a business — setting strategy, managing customers, making trade-offs between investment and margin. The distinction is not merely organizational. It is the difference between understanding how decisions are made and actually making them.
Then, in late 2011, AMD came calling. The company was being run by Rory Read, a former Lenovo executive who had been brought in as a turnaround specialist after the forced resignation of his predecessor, Dirk Meyer. Read — a crisp operational manager with a background in services and supply chain, not chip design — recognized that AMD needed product leadership and engineering credibility, and Su had both. On December 5, 2011, AMD extended Su an offer letter for the position of Senior Vice President and General Manager of the Product Group, reporting directly to Read, at an annualized base salary of $575,000 with a conditional sign-on bonus of up to $225,000. The offer letter, preserved in an SEC filing, is a remarkable artifact — mundane in its bureaucratic details (all new AMD employees start on a Monday), revealing in what it omits: any mention of the existential crisis Su was walking into.
She joined AMD in January 2012. The vast majority of the company's revenue came from chips for personal computers — a market that was contracting as smartphones devoured the consumer electronics landscape. AMD was bleeding money. Its products lagged Intel's by a generation or more. The company's GPU business was a distant second to Nvidia. But there were assets. AMD held a permanent, irrevocable license to the x86 instruction set architecture, one of only two companies in the world (the other being Intel) with that privilege. It had deep engineering talent, demoralized but not gone. And it had recently won design contracts to supply semi-custom chips for Sony's PlayStation 4 and Microsoft's Xbox One — a business Su had helped secure, and one that would prove crucial in keeping the company alive while she rebuilt everything else.
I felt like I was in training for the opportunity to do something meaningful in the semiconductor industry. And AMD was my shot.
Within two years, Su was promoted to chief operating officer, overseeing all business units, sales, and global operations. Then, on October 8, 2014, AMD announced that Read was stepping aside and Su would become president and CEO. She was forty-four years old, the first woman to lead AMD since its founding in 1969, and — in a statistic that is both a milestone and an indictment — one of the only female CEOs of a major semiconductor company in the world.
Three Bets, Five Years
Su's first weeks as CEO were consumed not by grand strategic pronouncements but by something more fundamental and more difficult: listening. She traveled to AMD facilities around the world, hosting town hall meetings with employees, collecting ideas, hearing suggestions, and — most importantly — addressing anxieties. The company was terrified. People had watched colleagues get laid off. They had watched product launches fail. They had watched the stock price crater. What they needed from a new CEO was not inspiration but credibility, and Su's engineering pedigree provided that in a way that no amount of charisma could have.
What emerged from those weeks was a strategy so simple it could fit on an index card: create great products, deepen customer trust, simplify operations. Three priorities. Three bets. The simplicity was the point. AMD had been diffusing its energy across too many mediocre initiatives, chasing every market without dominating any. Su's genius was subtraction — the willingness to identify what AMD should stop doing in order to concentrate its limited resources on what it could be best at.
"It is really important when you're a technology company to decide what you are really, really good at because you have to be the best, number one or number two," she told CNN. "It's all about choosing those battles." The echo of Jack Welch is intentional or coincidental, but the application was distinctive. Su didn't just choose which markets to enter. She chose which product bets to make within those markets, and she chose to make them on a five-year time horizon that was, by the standards of a company fighting for quarterly survival, almost recklessly long.
The critical bet was Zen. AMD's existing CPU architecture had fallen hopelessly behind Intel's, and no amount of incremental improvement would close the gap. Su authorized a ground-up redesign — a new chip architecture that would take years to develop, years more to produce, and that would either restore AMD's competitiveness in the CPU market or consume resources the company could not afford to waste. She told the board and shareholders it would take five years. The board, desperate for a credible plan, agreed. The shareholders, many of whom were measuring their positions in quarters, not years, were less patient.
"The most important thing for us back in 2014 was, first, choosing the right markets that were good for technology," Su said. "It's not what you do today but it's really what you do year after year, after year that people pay attention to." This is the voice of an engineer who understands compounding — not financial compounding, though that would come, but technological compounding: the way a sequence of correct decisions, executed with discipline over multiple product cycles, creates a gap between you and your competitors that eventually becomes unclosable.
The Architecture of Zen
The Zen architecture, when it finally arrived in 2017 in the form of AMD's Ryzen desktop processors, was a vindication so complete that it reshaped the entire competitive landscape of the semiconductor industry. For years, Intel had operated with something approaching complacency in the x86 processor market — a monopoly in all but name, with AMD serving as a nominal competitor whose products were tolerated by the market mainly because antitrust regulators preferred having two sources. Zen changed that. The new chips were not merely competitive with Intel's offerings. In many benchmarks, they were superior — faster, more power-efficient, and priced aggressively enough to force Intel into a pricing war it had never expected to fight.
The server variant, EPYC, followed, and it was EPYC that would transform AMD's financial trajectory. Data centers — the vast warehouses of servers that power Google, Amazon, Microsoft, and Meta — represent the most profitable and strategically important segment of the semiconductor market. When Su took over, AMD's share of data-center CPUs was functionally zero. By 2025, it had climbed above 40%. The math is staggering: each percentage point of data-center market share represents billions of dollars in annual revenue, and AMD had gained forty of them.
The key to Zen's success was not just the architecture itself but the manufacturing strategy behind it. AMD had spun off its foundry business in 2009, becoming a "fabless" chip designer — a company that designs chips but outsources their manufacturing to third parties, primarily Taiwan Semiconductor Manufacturing Company (TSMC). This had been viewed, at the time, as a desperate act of self-preservation. Su turned it into a competitive advantage. By partnering with TSMC — whose manufacturing technology was advancing faster than Intel's in-house fabs — AMD could ride TSMC's process leadership to deliver chips on cutting-edge nodes (7 nanometers, then 5, then 3) that Intel, stubbornly committed to its own struggling fabs, could not match.
"I don't personally use the terminology that
Moore's Law is dead," Su told Tom's Hardware. "The slow down in just pure manufacturing technology and process improvements has actually opened up new opportunities for innovation around architecture and design." The insight was characteristic: where others saw a problem (the deceleration of transistor scaling), Su saw an opening (the rising importance of architectural innovation, where AMD's engineering talent could compete without needing to outspend Intel on capital-intensive fabrication).
The Conviction to Wait
What distinguished Su's turnaround from the typical corporate rescue story — the cost-cutting, the restructuring, the activist-investor-driven shakeup — was its orientation toward the future. Most turnaround CEOs optimize for survival. Su optimized for dominance. She cut costs, yes — the previous administration had already done much of that painful work — but the core of her strategy was investment, not retrenchment. She was spending money AMD arguably didn't have on chip designs that wouldn't generate revenue for years.
This required a particular kind of nerve. AMD's quarterly earnings calls in 2015 and 2016 were exercises in managed desperation, with analysts probing whether the company could survive long enough for its product bets to pay off. Su's responses were measured, precise, and relentlessly forward-looking. She would acknowledge the challenges without dwelling on them, then pivot to the technology roadmap — a document she treated as almost sacred, because it was the thing that convinced customers to keep placing orders. "The most important thing," she told the Verge, is that "we had to demonstrate to customers that we had a technology roadmap that was compelling for three, five, even seven years."
This is the part of the story that rarely gets told in full because it is, by nature, uncinematic. Nothing is more boring than patience. Nothing is harder to dramatize than the decision to keep investing when the numbers say you shouldn't. But the technology roadmap — the sequential, year-by-year plan for improving chip performance, moving to smaller process nodes, introducing new features — was the central instrument of Su's turnaround. It was the thing she could show customers, partners, and investors that demonstrated AMD was not merely surviving but building. Roadmaps are promises. Su kept hers.
One of the things that I like to say about the semiconductor industry, or technology in general, is the decisions that we make today, you will really see the impact three to five years down the road. It is all about making the right bets.
By 2019, AMD was the best-performing stock in the S&P 500, up nearly 150% during the year. By 2020, its market capitalization had crossed $100 billion. By 2022, it surpassed Intel for the first time. The company that had been worth $2 billion when Su took over was now worth more than its century-old rival. Su's reaction, when asked if she felt vindicated, was characteristic in its flatness: "I don't know that I do that." Getting chips back from manufacturing and seeing they work as designed is her idea of something to celebrate. "It's super fun," she said, in the register of someone for whom the adjective "super fun" represents the upper bound of emotional display.
The Xilinx Wager and the Art of Acquisition
In October 2020, with AMD's stock riding high, Su made the biggest bet of her tenure: a $49 billion all-stock acquisition of Xilinx, the leading maker of field-programmable gate arrays (FPGAs) — reconfigurable chips used in telecommunications, defense, automotive, and data-center acceleration. It was the largest deal in semiconductor history at the time, and it was quintessentially Su: an acquisition driven not by empire-building but by a specific technical thesis about the future of computing.
FPGAs are, in simplified terms, chips that can be reprogrammed after manufacturing — flexible silicon that can be tailored to specific workloads in ways that fixed-function processors cannot. Su's bet was that the future of computing would not be one-size-fits-all but heterogeneous: different types of processors (CPUs, GPUs, FPGAs, custom accelerators) working together in orchestrated systems, with the value accruing to the company that could offer the most complete portfolio. Xilinx gave AMD a third leg, beyond CPUs and GPUs, and it brought with it a customer base in embedded, automotive, and aerospace markets that AMD had never been able to reach.
The deal closed in February 2022, after a fourteen-month regulatory gauntlet that spanned the United States, European Union, United Kingdom, and China. Su was named chair of the combined company's board, adding governance authority to her operational control. AMD's expanded portfolio — CPUs, GPUs, FPGAs, and adaptive computing — was now the broadest in the industry, a deliberate contrast to Nvidia's GPU-centric strategy and Intel's vertically integrated model.
A smaller but strategically revealing acquisition followed: the $1.9 billion purchase of Pensando Systems in May 2022, a data-center networking company whose distributed processing units (DPUs) gave AMD another piece of the data-center puzzle. Then came ZT Systems, a data-center infrastructure company, and Silo AI, the largest private AI lab in Europe. Each acquisition extended AMD's reach deeper into the AI ecosystem, layering software capability and systems integration on top of its silicon foundation.
The AI Surge and the Shadow of Nvidia
And then the world changed. On November 30, 2022, OpenAI released ChatGPT, and a technology that had been building quietly in research labs for years detonated into public consciousness. Within months, every major technology company was scrambling to build or buy the compute infrastructure needed to train and deploy large language models, and the demand for high-performance GPUs — the specialized chips that happen to be perfectly suited for the massively parallel computations required by neural networks — went from strong to insatiable.
Nvidia, which had spent two decades building its GPU ecosystem and its CUDA software platform, was positioned for this moment in a way no other company was. Its market capitalization, which had been roughly $360 billion at the start of 2023, would rocket past $3 trillion by mid-2024. Jensen Huang, Su's distant cousin, became the most celebrated CEO in technology. Of the $32 billion worth of AI data-center GPUs sold in the third quarter of 2024, Nvidia accounted for roughly 95%.
AMD was not caught flat-footed — Su had been investing in GPU compute for years, and the company's Instinct MI-series accelerators were genuine products, not vaporware — but the gap between AMD and Nvidia in AI chips was vast, and it was not primarily a hardware gap. It was a software gap. Nvidia's CUDA platform, a proprietary software layer that made it easy for developers to write code for Nvidia GPUs, had been built up over fifteen years and represented one of the deepest moats in technology. AMD's equivalent, ROCm, was years behind.
"I truly believe AI is the most transformational technology that I've seen in my career," Su told Bloomberg. "It's faster than anything we've ever seen." The enthusiasm was genuine but so was the challenge. AMD's data-center GPU business was growing fast — from essentially zero to a projected $5 billion in 2024 — but that number was dwarfed by Nvidia's revenues. The AI chip market, which Su predicted would reach $150 billion by 2027, was enormous enough for multiple winners, but winning required more than good hardware. It required an ecosystem.
Su responded with characteristic methodical aggression. She accelerated the Instinct GPU roadmap, launched the MI300X — which she called "the most advanced AI accelerator in the industry" — and committed to an annual cadence of new AI chip releases. She invested heavily in ROCm, hired AI software engineers, and in January 2025, after a scathing SemiAnalysis report detailed the software stack's deficiencies, AMD adopted what analysts called a "developer first" strategy, appointing Anush Elangovan as an "AI Software Czar" to lead external developer relations. She met personally with the SemiAnalysis team to discuss their findings.
"There are so few companies in the world that have access to the intellectual property that we have and the customer set that we have, and the opportunity frankly to really shape how AI is adopted across the world," Su told Fortune. "I feel like we have that opportunity." The statement is optimistic, but it is also a careful framing of AMD's position: not the leader, but one of the very few companies in the world with the capability to compete. In an industry where the barriers to entry are measured in billions of dollars and decades of accumulated IP, being one of three is not a consolation prize. It is a franchise.
The Geometry of Power
To understand Su's position as of late 2025, you have to understand the peculiar geometry of the semiconductor industry — a geometry she navigates with the spatial awareness of someone who has lived inside it for thirty years.
There are, broadly, three companies that matter in high-performance computing: Intel, Nvidia, and AMD. Intel, the oldest and largest, is struggling. It ousted CEO Pat Gelsinger in late 2024, its manufacturing technology has fallen behind TSMC's, and its market capitalization has shrunk to a fraction of AMD's. When Su was asked, at a Time event, who should replace Gelsinger, she declined to name names but noted it was "a tough job." The courtesy was genuine; the understatement was devastating.
Nvidia is the giant. Its CUDA ecosystem, its dominant market share in AI GPUs, and Jensen Huang's extraordinary capacity for strategic vision have made it arguably the most important company in technology. But Nvidia's dominance also creates a market dynamic that works in AMD's favor: customers — the Microsofts, Googles, Amazons, and Metas of the world — do not want a single-supplier dependency. They want optionality. They want a second source. This is not merely a preference. It is a structural imperative driven by procurement logic, supply-chain resilience, and antitrust caution.
AMD is that second source. But Su has spent her entire tenure trying to make AMD more than that — trying to change the company's identity from the perpetual runner-up to a genuine technology leader. "The history of the company is that we've been second source to other people," she told Fortune in 2015. "I think we really need to change that." The statement was aspirational then. By 2025, it was largely accomplished in CPUs, where AMD's EPYC processors are competitive with or superior to Intel's in most benchmarks, and partially accomplished in GPUs, where the Instinct MI series is a credible if distant challenger to Nvidia.
The geopolitical dimension has also elevated Su's profile. She testified before the Senate Commerce Committee in May 2025, alongside
Sam Altman of OpenAI, Microsoft President Brad Smith, and CoreWeave CEO Michael Intrator, making the case for maintaining U.S. leadership in AI. She negotiated an export deal with the Trump administration, alongside Huang, that carved AMD and Nvidia out of chip-export restrictions to China in exchange for a 15% cut of China sales to the U.S. government. The woman who once worried about keeping a $3 stock alive was now sitting across the table from senators and presidents, representing an industry that had become central to national security.
Weekend Meetings and Midnight Memos
The mythology of the relentless CEO has become, in Silicon Valley, something of a cliché — the founder who sleeps at the office, the executive who sends emails at 3 AM as a dominance display. Su's work ethic is intense, but it operates in a different register: less performative, more structural. She holds meetings on weekends with her senior team. She sends feedback on pre-read materials late at night, then calls managers in the morning to discuss them. (An AMD spokesperson clarified to Fortune that one widely reported anecdote about midnight calls was actually about a late-distributed pre-read for an early-morning meeting, with Su directing the team on which sections to focus on — a correction that somehow makes her sound more methodical, not less.)
"People are really motivated by ambitious goals," she told Time. "The previous strategy of, hey, let's just do a little bit better here and there — that's actually less motivational." The insight is counterintuitive but empirically sound: people would rather work punishingly hard toward a goal they believe in than coast toward incremental improvement. Su sets the bar high and holds people to it with a directness that former colleagues describe as both inspiring and unsparing. Patrick Moorhead, a tech industry analyst and former AMD executive, told Time that Su's high expectations can make it challenging for people to survive at AMD long-term. She smiled when asked about her reputation for zero tolerance for excuses. "Well, I like to win, if that's okay."
The winning, though, is inseparable from the engineering. Su still visits labs. She still looks at chip designs. She still gets, as she puts it, "super fun" excitement when chips come back from manufacturing and work as designed. This is not affectation. It is the residual identity of someone who, at the deepest level, is still the MIT PhD who took apart her brother's car to see the motor. The managerial layer — the strategy, the acquisitions, the Senate testimony — sits atop an engineering foundation that has never eroded, and it is this combination that gives her authority with the people who matter most in a semiconductor company: the engineers.
"I don't believe leaders are born," Su told Time. "I believe leaders are trained." The statement is an implicit autobiography. She trained herself — through the multiplication tables, the MIT PhD, the IBM apprenticeship, the Freescale P&L responsibility, the AMD crucible — and the training never stopped.
The Building with Her Name
In 2022, MIT named Building 12 — the home of MIT.nano, the Institute's state-of-the-art nanotechnology research facility — the Lisa T. Su Building. It is the kind of honor that, in academic circles, represents the highest form of institutional recognition, reserved for those whose contributions to a field are considered permanent. Su had previously established the Lisa Su Fellowship Fund in 2018 to support female graduate students in nanotechnology, and she served on the Electrical Engineering and Computer Science Visiting Committee for a decade. In December 2025, MIT announced that she would deliver the 2026 Commencement address.
The accolades have accumulated with the relentlessness of compound interest: Time CEO of the Year, Fortune Businessperson of the Year, Forbes billionaire list (net worth estimated at $1.3 billion as of April 2024), the Robert N. Noyce Medal (the semiconductor industry's highest honor), the IEEE Fellow designation, the Franklin Institute's Bower Award for Business Leadership, membership in both the American Academy of Arts and Sciences and the National Academy of Engineering. She is the highest-paid female CEO in America for five years running, according to an Associated Press survey. She is one of approximately fifty-two women leading Fortune 500 companies, in a semiconductor industry where women account for just 30% of the workforce.
The numbers are stark. The achievement is undeniable. And yet the most revealing thing about Su's relationship to all of it might be what she told Time's editor-in-chief at a dinner in New York, when he asked her to share an example of the most important kind of decision — the decision not to do something. Her board, she said, sometimes asks why AMD isn't making phones or tablets or other popular consumer devices. Her answer: AMD has to stick to what it does best. Making the underlying technology. "Big computers were not sexy 10 years ago," she said, "but they're kind of sexy now."
The moderator responded: "Well, Lisa Su, thank you for making supercomputers sexy."
She might have laughed. The sources don't say. But somewhere in a lab in Santa Clara, an AMD chip was coming back from TSMC, and all 40 billion of its transistors were working exactly as designed, and that — for the woman who took apart her brother's car to see the motor, who chose electrical engineering because it was the hardest major, who walked into a $3 company and saw not wreckage but possibility — that was the real thing. That was super fun.
