Right now, Marvell’s chips are humming behind your day.

Their chips are everywhere you don’t see them, powering the networks and infrastructure behind your daily life.

They help move data inside Amazon’s cloud, connect the multi-gigabit fiber boxes for your home internet, and link the high-speed ethernet networks inside modern cars.

Open up the latest hard drives storing your cloud files, or enterprise network equipment in a Fortune 500 data center, and you'll find Marvell at work.

With over a billion chips shipped each year and a market cap of $74B, Marvell quietly powers the backbone of our connected digital world.

But it all started in 1970s Jakarta with a Chinese-Indonesian boy sitting in his parents' auto parts store.

‍

🔌 The boy who only saw electronics

In 1970s Jakarta, power cuts came without warning. Most families didn't own a telephone, much less a computer.

But thirteen-year-old Sehat Sutardja saw something else.

He sat in his parents' auto parts store with a broken radio spread before him. He read the names stamped on tiny components. Fairchild, Texas Instruments, RCA. American names that whispered of another world.

While neighborhood kids played football in the streets, Sehat hunched over circuit boards.

He often visited his brother Pantas in Singapore. Together, they shared a passion for building and experimenting with electronics, sometimes tackling projects inspired by what they saw in science magazines or textbooks.

He saw the joy of making things work with his own hands. Watching a homemade device spring to life was a feeling that stayed with him for life

By thirteen, Sehat had his radio repair certification and he was repairing radios in his free time.

Even until his old age, he kept that license in his wallet through everything that followed as a reminder for his humble beginnings and his passion for electronics.

‍

💡 The man who kept digging

That obsession carried him from Jakarta to Iowa State University in the US, then UC Berkeley for his master's and PhD. At Berkeley, he met Weili Dai (who later became his wife in 1985) in a campus elevator.

At Berkeley, Sehat stood out.

It was not for how much he knew, but for how far he'd go to understand what others took for granted.

In Professor Stephen Lewis's class, most students solved problems the standard way. Sehat kept digging until he found something better. He loved to disregard conventional wisdom in pursuit of something better.

At his first job at Micro Linear Corp, they handed him the company's design library. It had thousands of pre-built components, and they were tested and proven.

Sehat redesigned the circuits from scratch, slashing disk controller latency from 100 nanoseconds to 10. It was a 10x improvement that seemed impossible.

His managers didn't believe it would work, but the circuits ended up working.

Sehat had done it again. He proved that the better solutions live just past where most people stop looking.

‍

🏠 The kitchen table bet

By the mid-90s, Sehat had spent years designing chips for other companies. But his wife, Weili, had encouraged him to start something of their own.

So in early 1995, Sehat was ready to build his own path. Along with his wife as a co-founder, he asked his brother, Pantas, who had just left IBM's Almaden Research Center, to join them.

Together, they incorporated Marvell Technology Group in 1995, inspired by the word “marvelous” because of the marvelous global company they wanted to build. Oh, and because successful tech companies like Intel, Novell, and Nortel ended in "el" or "ell."

They bootstrapped the company with $350,000 raised from their family.

The stakes were high. Sehat and Weili had two young children depending on them, and college tuition was looming years ahead.

And they were building in a market dominated by semiconductor giants with billion-dollar R&D budgets and decades of market leadership.

If Marvell failed, the blow would be huge to them and their family. There was no safety net and their children may pay the price for their parents' dream.

Still, the family bet everything. There was no plan B.

‍

🚫 Nobody believed them

They were using CMOS (Complementary Metal-Oxide-Semiconductor) technology to build chips for hard drives.  

At the time, Texas Instruments dominated the field with bipolar technology. The company had the resources, talent, and R&D budget to explore every option. For speed-critical applications like hard drive read channels, bipolar had always been the standard.

CMOS was considered too slow.  

That wasn’t just one company’s opinion. It was the industry consensus, backed by billions in research and decades of experience.  

The Sutardjas believed one of the world’s most powerful semiconductor companies was wrong. More than that, they were betting their family’s future on it.  

They took their prototypes to industry conferences, hoping to spark interest. Instead, they faced polite skepticism and subtle dismissal. The technology had been evaluated. The verdict was in.

But Sehat had seen something the entire industry missed. Everyone said you needed bipolar technology for the speed hard drives required.

Sehat disagreed. What if you stopped trying to force CMOS to work like bipolar? What if you designed everything around CMOS's natural strengths—lower power consumption, smaller die size, cheaper manufacturing—instead?

You'd get speed, efficiency, and performance at lower cost.

The doubt didn’t go away.

After each rejection, he'd revisit his numbers, checking for what he might have missed. To him, the logic made sense. Yet, everyone around him seemed convinced he was chasing the impossible.

If he was right, they'd redefine an industry.

‍

🔧 The chaos no one sees

Even after Marvell incorporated in 1995, it was far from a polished operation. They had no office, no HR, and no formal supply chain.

For the first few years, none of the three founders drew a salary. Every dollar went into prototypes, parts, hiring their first four engineers, and staying afloat.

Rent, diapers for their two young sons, and groceries—all these things came from savings and the money their family entrusted to support them.

Their first prototypes were built on a shoestring. They were designed late at night by Sehat at home, with parts assembled on a makeshift workbench.

Sehat focused on design, Pantas on implementation and operations, and Weili on business development, cold outreach, and anything else that needed doing.

Sehat even wrote software to model chip behavior by hand, since commercial tools were too expensive. They couldn't afford layout engineers, so he sketched schematics himself. Weili fielded calls and secured meetings with potential customers, even while managing a young household.

The pressure was constant. Every month without revenue meant burning through what little they had left. If the chips failed, it was possible that there would be no second chance.

‍

🎯 The door that wouldn't open

By the end of 1995, Marvell had what many thought was impossible: a working read channel prototype using CMOS. It was fast, cheap, and consumed less power than anything on the market. On paper, it should've changed everything.

But no one wanted it.

They sent samples to three major hard drive companies. Each confirmed the same thing: the chip outperformed expectations. The technology worked. Still, the answer was no. Again and again and again.

Why? Because it came from a company no one knew, with no track record, no factory, and no investors.

They were too small and too risky.

The customers weren't rejecting the chip. Instead, they were rejecting them. The performance didn't matter if Marvell couldn't guarantee supply, support, or survival.

Weili, who ran sales while raising two young boys, worked the phones every day. She chased every lead, called every contact.

There were no buyers and no revenue. Even if they finally proved they made it work.

‍

🎲 The benchmark

Finally, one cold call connected. Weili reached Kenneth Burns, a scientist at Seagate's Minnesota office, who was struggling with a chip from Texas Instruments that couldn't keep pace with the next generation of drives.

Burns explained that Seagate's upcoming drives would require faster chips than anything built so far. He gave Marvell a benchmark to hit, which was a speed target that seemed nearly impossible.

Sehat's brother recalls that their chips were already running close to the benchmark, but close wasn't good enough.

Weili promised Seagate that Marvell could do better. The stakes were clear: hit the benchmark and Seagate might take a chance on them. Miss it, and it was over.

For three months, Sehat refined the design. Every simulation mattered. Every optimization brought them closer—or revealed another constraint. The prototype either worked or their credibility was gone.

And the Marvell team hit the mark.

Seagate became their first major customer, and that milestone unlocked everything else. The chip that industry leaders had called impossible sailed into production. Better yet: Seagate decided to drop Texas Instruments entirely and went all in with Marvell.

The founders who'd been betting their family's future in a kitchen suddenly weren't so crazy anymore.

‍

💰 The breakthrough

After Seagate signed production orders, the industry took notice. Samsung called. Then Western Digital.

Word spread fast through the tight-knit community of hard drive manufacturers. Within months, companies were approaching this unknown startup in Santa Clara. The same company most had ignored just months earlier.

But validation didn't mean survival. Revenue was trickling in, but not fast enough to scale production.

They needed engineers, equipment, and manufacturing capacity. Every new customer order meant finding cash to fulfill it. The gap between success and collapse was still there.

Then Filipino chip entrepreneur Dado Banatao, who had seen their progress, offered them $1M for R&D.

For the first time since founding Marvell, the three founders could exhale.

‍

🏢 Building relationships and trust

Manufacturing became its own nightmare.

Marvell didn't own fabs (fabrication facilities specialized to manufacture semiconductor chips). They designed chips that foundries in Taiwan would manufacture.

But foundries prioritized big customers with proven track records and consistent volume. Marvell's orders kept getting bumped.

As a tiny startup with no leverage, they needed to build credibility with manufacturing partners.

Sehat's brother Pantas spent months cultivating relationships at TSMC, the foundry that would manufacture their chips. Carefully chosen gifts, meeting after meeting, and learning the unspoken language of Taiwanese business culture.

It wasn't about one meeting or one deal. It was about proving Marvell would be around next quarter, next year. To signal that their orders would grow and the relationship would be worth prioritizing.

Slowly, TSMC began treating them less like a little startup and more like a future partner.

Orders stopped getting bumped and production timelines became reliable. Now, the chips that Sehat designed could finally reach customers at scale.

‍

🚀 Building an empire

By 2000, Texas Instruments made a strategic decision to exit the read-channel semiconductor sector, a market they once dominated.

Having initially dismissed CMOS technology as too slow, TI later acknowledged they were "frankly late in the transition to digital CMOS" and faced significant "execution issues" as the industry rapidly shifted from bipolar to CMOS designs.

During this pivotal transition, Marvell's market share in read channels grew from just 2-3% in 1998 to 11% in 1999, largely capturing the business Texas Instruments vacated as the industry moved almost overnight toward CMOS technology.

In June 2000, Marvell went public on NASDAQ. The stock was offered at $15 per share and closed the first day near $56.63.

At the time, they had raised about $90M and were already profitable.

But Sehat wasn't satisfied dominating storage. He saw the future, and it was wireless and mobile. That meant every device would need chips to connect, communicate, and process.

In 1998, Marvell opened a research and development center in Israel. In 2003, the company acquired RADLAN to expand its networking technology portfolio. In 2006, Marvell bought Intel’s XScale processor business—marking a major expansion into mobile and wireless chips, and paving the way for deeper collaborations with global tech giants.

When Apple launched its original iPhone in 2007, it featured a Marvell Wi-Fi chip, marking a milestone for the company’s presence in mobile devices.

By the late 2000s, Marvell silicon could be found everywhere: Google’s Chromecast ran on Marvell SoCs, and e-readers, game consoles, and smartphones routinely featured Marvell chips. The company that started in a kitchen now powered the hidden infrastructure of the digital world.

By this point, Sehat Sutardja had accumulated over 440 patents, and earned recognition as Inventor of the Year by the Silicon Valley Intellectual Property Law Association.

In 2009, he decided to invest in the next generation of engineers at his alma mater. Sehat, Pantas, and Weili donated more than $20M to UC Berkeley for research and education.

The university named a building in their honor, the Sutardja Dai Hall, a 141,000-square-foot research facility where hundreds of students and faculty now work to advance and innovate technology.

‍

🔮 Scrutiny in success

Success brought visibility. And with visibility came scrutiny.

In 2006, the SEC investigated Marvell for backdating stock options—choosing grant dates with hindsight to make them more valuable. The investigation dragged on for two years.

In 2008, Marvell paid $10M to settle. Weili stepped down as COO. Sehat remained as CEO, but the company they'd built from nothing now carried a stain.

The issues didn't end there. In 2016, activist investor Starboard Value took a 7% stake in Marvell and pushed for change.

A board investigation followed, examining how management ran the company they'd founded. The findings: sales teams had been pressured to meet revenue targets, with some revenue booked prematurely. No fraud. But the internal controls hadn't been followed properly.

After months of boardroom battles, the conclusion came: "the time has come to move in a new leadership direction."

On April 5, 2016, Sehat and Weili stepped down from their leadership roles. Twenty-one years—from kitchen table prototypes to a multi-billion dollar company—ended with a board decision. The founders who had been right when the entire industry said they were wrong were pushed out by the company they'd created.

‍

💙 The legacy lives on

After leaving Marvell, Sehat didn't stop building. In 2021, he co-founded Silicon Box with Weili, focusing on chiplet packaging for AI. They opened a $2B facility in Singapore in 2023 and raised a $200M Series B around January 2024, hitting unicorn status.

But shortly after, on September 18, 2024, Sehat Sutardja died suddenly in Las Vegas at age 63.

Today, Marvell ships over a billion chips annually. More than 6,500 employees worldwide. The kid who fixed radios didn't just build a company. He proved that deep technical conviction can reshape entire industries.

And the new company Sehat co-founded with Weili, Silicon Box, is still around today. They recently announced they shipped 100M units from their flagship factory in Singapore.

Walk through UC Berkeley and you'll pass Sutardja Dai Hall. Inside, students hunched over circuit boards chasing the same curiosity that once drove a boy in Jakarta.

Sometimes the boldest bet is believing in what you know to be true, even when everyone else says it can't be done.

‍

Takeaway Scoop

Lessons founders and builders can learn from the Marvell Technologies story.

• Technical disruption happens at the architecture level, not component level. Everyone tried to make CMOS faster to compete with bipolar. Sehat didn't optimize the existing approach; he invented a new one. When industries say "it can't be done," they usually mean "it can't be done our way."
• Don't forget the "proof-to-revenue" gap. It could kill you. Marvell had working chips by late 1995 that beat everything on the market. But for two full years, nobody would buy them. Why? Hard drive companies needed guaranteed supply chains, 24/7 support, and certainty you'd exist next quarter. A startup with 3 founders and 4 engineers couldn't provide that, no matter how good their chips were. Budget for this purgatory.
• Your first customer can determine your market position. Seagate didn't just buy chips. They dropped Texas Instruments entirely for Marvell. This wasn't a trial order; it was a change that made every other hard drive maker take notice. That's why Sehat spent three months perfecting their chip to hit Seagate's "impossible" benchmark. In some industries, your first major customer becomes your calling card. Pick the one whose endorsement rewrites the industry narrative.
• Your supply chain relationships are your capacity. Marvell never built a fab (production facility), yet dominated the market. How? Pantas spent months at TSMC in Taiwan—learning the culture, choosing the right gifts, proving Marvell would grow. This got their orders prioritized over bigger companies. Your competitor has the same salesperson's number. The difference is whether your orders get bumped when capacity gets tight.

‍