“Goddess” Su Ma

After the meme with the wrong face went viral, I dug deep into Lisa Su’s early years. Conclusion: it may differ from the popular version.

Let me start with a bucket of cold water.

Most articles about Lisa Su follow a near-identical template: an immigrant girl from Tainan, South Taiwan, who worked hard and achieved legendary status, took over a dying AMD at a critical moment, pushed the stock from a couple of dollars to hundreds, took on Intel, and chased NVIDIA — a complete underdog success story.

This version isn’t exactly wrong, but it strips away her most valuable asset.

Because if “hard work + elite school + critical mission” could explain a person, Silicon Valley would have a hundred Su Mas. What’s truly rare is a judgment she formed over the two decades when nobody knew the term “Su Ma” and she was just an engineer working on device physics — a habit of picking problems that are hard, concrete, and deliverable, while avoiding sexy but flashy trends. AMD’s turnaround was the payoff of that judgment reaching age fifty, not a stroke of luck one day.

So this piece focuses only on the early years. Everyone knows the CEO story; I actually think it’s less important.

1. From Tainan to Queens: A Standard STEM Assembly Line

Born in November 1969 in Tainan, Taiwan, she emigrated to the US with her parents at age 3 and grew up in Queens, New York.

Hardcore family setup: father was a statistician who started quizzing her with multiplication tables at age 7; mother was an accountant who later started her own business, instilling business concepts from a young age. The “permitted careers” at home were only three — doctor, pianist, engineer. At age 10, she started dismantling her brother’s remote-control car to fix it; in junior high, she got her first computer, an Apple II.

Before you get too emotional, hold on.

This is actually a replicable template for middle-class Asian immigrant engineer families, not some genius myth. The same script plays out in Silicon Valley, Hsinchu, and Zhongguancun every year in large numbers. Calling it “a destined chosen one” is lazy.

What’s worth pausing on are two details:

First, she had a physiological preference for “taking things apart to see how they work.” This isn’t just inspirational — it’s the raw material of a device physicist. What she later did at IBM was essentially the same thing, scaled down to nanometers.

Second, and nine out of ten versions skip this: the accountant mom teaching her about business. This thread is extremely important because it explains why she never became a “pure tech nerd.” We’ll come back to this clue in Section 4.

2. Eight Years at MIT: Actively Choosing a Cold Bench

In 1986, she graduated from Bronx High School of Science — New York’s top public high school — and went straight to MIT for a combined BS/MS/PhD, earning her doctorate in 1994, all in electrical engineering.

Her PhD thesis was on SOI (Silicon-On-Insulator) devices — specifically, extremely scaled SOI MOSFETs.

This choice itself reflects taste.

SOI was a cold, low-level direction back then; the industry hadn’t even verified whether the path was viable. It wasn’t the kind of hot topic that makes news or lets a professor show off in the department. But it solved a real problem — how to control leakage and parasitic effects in ever-shrinking transistors. The two early technical achievements she’s remembered for (SOI and copper interconnect) are both exactly this type of “hard, dirty, but unavoidable” tough nut.

My judgment is simple: From her student days, she never chased trends. Others picked topics based on “how trendy,” she picked based on “how hard and how real.” This habit is far more important than how many degrees she earned.

【Image 1】1994, PhD graduation from MIT at age 24. Thesis direction: extremely scaled SOI MOSFETs — one of the coldest and hardest tracks of that era.

3. Twelve Years at IBM: The Battle of Copper Interconnect

She first briefly landed at Texas Instruments (TI) in 1994, then joined IBM in 1995, staying for twelve years. This is where her early legend truly took shape.

Her masterpiece was copper interconnect.

Quick background: For a long time, the metal used for wiring inside chips was aluminum. Aluminum has physical limits; as you scale down and speed up, aluminum can’t keep up. Everyone knew copper was better — lower resistance, higher current capacity — but there was a nearly insurmountable hurdle called “copper poisoning”: once copper atoms diffused into silicon, at best it would ruin that chip; if it contaminated the equipment, the entire production line would suffer.

This was a dirty, hard, and not-at-all-sexy engineering problem. But precisely because it was so hard, whoever solved it first would own a generation of technology.

Her team’s job was to suppress this contamination problem and create a set of “recipes” that could actually run on the production line. One of the most counterintuitive findings: they discovered that trace impurities in copper weren’t harmful but actually significantly improved reliability. Cyprian Uzoh, a chemist on the team, joked: “Just add a little salt and pepper, it doesn’t hurt anyone.”

Let me align the timeline for you so you don’t have to look it up: September 1997, IBM announced a manufacturable copper process; 1998, the first copper interconnect chips shipped. The contemporary PowerPC went from 300 MHz to 400 MHz — a 33% speed boost (strictly speaking, a combination of copper and new processes; copper alone contributed about 15%). Then it became an industry standard — and continues to be used today.

Remember the shape of this: a concrete, deliverable-to-factory, industry-adopted breakthrough. Not a vision on a PPT, not a demo — something that actually went into every chip and has run for nearly thirty years.

If there’s only one sentence worth taking from this article, it’s here — the kind of problem an engineer should truly focus on.

4. The Severely Underestimated Move: She Actively Shifted Toward Business

In 2000, she did something outsiders wouldn’t understand but insiders would immediately recognize: she became Technical Special Assistant to then-IBM CEO Lou Gerstner for one year.

Outsiders might see this as gilding the lily or networking. Insiders know this move was her taking a course — how to turn technology into a business.

At the CEO’s side, she saw up close not code, but how a giant company makes decisions, allocates resources, and cuts seven out of ten equally important projects. You can’t learn that in a lifetime spent in the lab.

After that, she founded IBM’s Emerging Products division — starting with just herself, essentially running a startup inside IBM. She later grew the team to about a dozen people, working on low-power chips, biochips, and other “emerging products”; she also participated in the joint IBM + Sony + Toshiba Cell processor development — later used in the PS3.

【Image 2】2005, holding the IBM/Sony/Toshiba co-developed Cell processor. The following year, she was promoted to Vice President of IBM’s Semiconductor Research & Development Center, one of the highest-ranking female executives in the US chip industry at the time.

Now, bring back the “accountant mom” from Section 1.

Many technical experts are forever stuck on “only know technology.” They can solve world-class problems but never understand money, customers, and trade-offs. What makes Su Ma different isn’t that she’s technically stronger, but that she actively chose to fill the business gap in the middle of her career. From having her mom teach her business as a child to being a CEO’s assistant in mid-career — this “technology × business” dual track is the real fork in the road separating her from a “lifetime senior engineer.”

【Image 3】IBM period, photo with then-CEO Lou Gerstner. Those years as his Technical Special Assistant, she learned “how to turn technology into a business.”

5. The Rest of the Story: Just Payoff

The rest is familiar, so I’ll fast-forward.

In 2007, she moved to Freescale as CTO; in 2012, she joined AMD; in 2014, she became CEO — the company was near bankruptcy, with a stock price around two or three dollars. She bet on a completely new Zen architecture, created Ryzen and EPYC, took on Intel head-on, and is chasing NVIDIA on AI/GPU. As of writing (June 2026), the stock is around $450 — over a hundred times what it was when she took over.

But what I want to say is precisely: this part isn’t as worth telling.

Because it’s merely the payoff of the previous twenty years. The problems she chose after entering AMD are the same type as the ones she chose in her early twenties (SOI) and late twenties (copper contamination) — hard (rebuilding CPU architecture from scratch), concrete (Zen), and deliverable (consistent shipment schedule). The pattern is identical.

In other words, the critical-mission story of 2014 only worked because someone between 1994 and 2007 had turned judgment into muscle memory.

Conclusion: A Non-Inspirational Takeaway

I’m not going to give you a “hard work leads to legend” conclusion, because that’s false.

What you can truly extract from her early years, and still valid to use, is this:

When no one is calling you “X Ma” or “X God” yet — whether you bet on trends, or on hard, concrete, deliverable problems — determines whether you’ll have cards to play twenty years later.

Copper interconnect was developed nearly three decades ago, and it’s still running inside every chip today. Trends don’t.

This point is especially applicable to people working in AI and engineering — the tool you chase today will be obsolete next month; but a hard problem you actually solved won’t.

As for the mismatched-face meme at the beginning, it ironically serves as a reminder: if you see her as a “goddess,” you learn nothing; if you see her as a series of choices, that’s where you find what’s truly valuable.

🥚 Easter Egg: Lisa Su and Jensen Huang (old Huang) are relatives. Jensen’s mother is the younger sister of Lisa’s maternal grandfather; by generation, Jensen is Lisa’s “maternal uncle” (in Western terms, first cousin once removed, though Jensen is a few years older). But they didn’t grow up together; they first met at an industry event only after each became famous. She later said in a Bloomberg interview: “No family dinners.” Two people from Tainan, leading NVIDIA and AMD respectively, sitting across from each other at the AI chip table — this story itself is better than any formulaic success story.

Sources: Wikipedia, MIT Technology Review, IBM Research, Carnegie Corporation, EE Times, CNN, Fortune, Bloomberg. Early technical details (SOI / copper interconnect / timeline) follow MIT Technology Review and IBM Research official versions; AMD stock price data is from June 2026.