AMD: The Comeback That Should Have Been a Bankruptcy

In 2015, almost nobody believed Advanced Micro Devices would survive. The company’s stock had collapsed toward two dollars a share. It carried about two billion dollars in debt. Its latest chip architecture, code-named Bulldozer, had been a flop, losing badly to rival Intel on both performance and efficiency. For decades AMD had been the perpetual second place in computer chips, the scrappy understudy to the giant up the road, and now the obituaries were being written. Analysts modeled bankruptcy. The market had effectively given up.

Into that wreckage walked a soft-spoken engineer with a doctorate from MIT named Lisa Su, who had taken over as chief executive in late 2014. She inherited a company with almost no margin for error and one realistic shot at survival: a clean-sheet processor architecture, years from shipping, that would have to beat Intel to matter at all. She bet the company on it.

What happened next is one of the greatest comebacks in the history of business. This is the story of the eternal number two, the company that came within a breath of bankruptcy, and then out-executed a complacent giant so completely that, a decade later, its data-center business would pass the very rival that had dominated it for forty years.

The Eternal Number Two

AMD was founded on the first of May, 1969, in Sunnyvale, California, by Jerry Sanders and seven colleagues from Fairchild Semiconductor. Sanders was a flamboyant salesman, famous for lavish parties and the conviction that a chip company had to be as bold in marketing as in engineering. From the beginning, AMD lived in the shadow of a company founded a year earlier by other Fairchild alumni, just up the road: Intel.

For much of its life, AMD was a second source, a maker of chips compatible with Intel’s, licensed to produce the x86 processors that ran the world’s PCs. The relationship was a rivalry and a dependency at once. AMD spent decades as the cheaper alternative, occasionally landing a real blow. Its finest early hour came in 2003, when it pioneered 64-bit computing for the x86 standard with its Opteron and Athlon 64 chips, beating Intel so decisively that Intel was forced to adopt AMD’s own 64-bit extensions. For a brief, glorious moment, the understudy led the star. It would not last.

The Years of Dying

In 2006, AMD made a fateful bet, buying the graphics-chip company ATI for about five and a half billion dollars. The logic, combining processors and graphics, was sound, but the price was steep and the debt was heavy. Almost immediately the integration strained the company, even as Intel’s new Core processors pulled decisively ahead. AMD’s market share eroded, its losses mounted, and the ATI debt became an anchor.

To survive, AMD made a decision that looked like surrender and turned out to be salvation. In 2008 and 2009 it spun off its own factories, the expensive, capital-devouring fabrication plants, into a separate company called GlobalFoundries, funded in part by an investment arm of Abu Dhabi. AMD would no longer manufacture its own chips. It would only design them, and pay someone else to build them. For a chipmaker, this was once heresy. AMD’s own founder had famously declared that real men have fabs. But owning fabs was exactly what AMD could no longer afford. The semi-custom chips it won for the PlayStation 4 and Xbox One in 2013 brought in just enough cash to keep the lights on. By the mid-2010s, with Bulldozer failing and the debt still heavy, the company was on the edge of the grave.

Lisa Su and the Zen Bet

Lisa Su’s strategy was deceptively simple and brutally disciplined. She declared that AMD would compete on high performance, not on being the cheap option, and she killed the distractions that were not core to that goal. Everything went into a new, clean-sheet processor architecture called Zen, and a clever design philosophy called chiplets, building processors out of smaller modules stitched together, which yielded better and cost less than the giant monolithic chips of the old way.

Crucially, AMD no longer had to manufacture those chips itself. It could design Zen and then have it built by TSMC, the Taiwanese foundry that had seized the manufacturing lead from a stumbling Intel. AMD was now a nimble designer riding the best factory in the world, while its giant rival was weighed down by its own struggling fabs. When the first Zen-based Ryzen processors for PCs and EPYC processors for servers launched in 2017, they were genuinely competitive, and then genuinely superior. The understudy had a hit, and it began taking back the share it had lost over a generation.

Decision Point

What Everyone Gets Wrong

Mistake #1: Believing AMD beat Intel by outspending it. The opposite is true. Reality: AMD’s research budget was a fraction of Intel’s for years. It won by out-designing its rival, with the chiplet approach, and by renting the world’s best manufacturing rather than trying to build it.

Mistake #2: Reading the fabless move as a retreat. Giving up your own factories sounds like weakness. Reality: Spinning off manufacturing was the move that freed AMD to win. It let AMD ride TSMC’s process leadership while Intel was dragged down by its own struggling fabs.

Mistake #3: Crediting the comeback purely to AMD’s brilliance. The turnaround was real, but it had help. Reality: AMD executed superbly, and Intel wounded itself with missed transitions and a lost manufacturing lead. The comeback was both AMD’s discipline and its rival’s self-inflicted decline.

Mistake #4: Concluding AMD has won. The narrative tempts a victory lap. Reality: AMD is still the challenger, far smaller than Nvidia in artificial intelligence, and the chip business is savagely cyclical. The same company that nearly died once is not immune to dying again.

How the Underdog Kept Climbing

Zen was not a one-time miracle but the start of a relentless cadence. AMD shipped a new, improved architecture year after year, taking share in PCs and then, more lucratively, in the data center, where its EPYC server chips steadily ate into Intel’s most profitable market. In 2022 came a milestone few would have believed in 2015: AMD’s market value surpassed Intel’s for the first time in history. That same year AMD completed its roughly forty-nine-billion-dollar acquisition of Xilinx, pushing into adaptive and embedded computing.

Then came artificial intelligence, the largest opportunity in the history of the industry, and AMD pivoted hard toward it. Its Instinct accelerators, the MI300 and its successors, became the leading credible alternative to Nvidia’s dominant AI chips, and the giant cloud companies, hungry for any second source, came calling. The comeback that began with a single desperate bet on Zen had carried AMD from the edge of bankruptcy into the center of the most important technology race on Earth.

The Honest Present

By 2025 the transformation was complete enough to be measured in the rawest possible terms. AMD’s revenue reached the mid-thirty-billions of dollars, up sharply, with the data center now more than half of its business. In one quarter that year, AMD’s data center revenue overtook the entire data center group of Intel, the rival that had dominated it for four decades, for the first time. Its market value, once near two billion dollars, ran into the hundreds of billions. Lisa Su, who had taken over a company being measured for its coffin, was named TIME’s chief executive of the year for 2024.

And yet the honest present is not a coronation. In artificial intelligence, AMD remains the clear underdog to Nvidia, fighting for a second-source role rather than the lead. Its fortunes are tied to TSMC’s factories and to a chip cycle that has humbled it before. The lesson of AMD’s own history, that a complacent giant can be toppled by a leaner, hungrier designer, is precisely the lesson a future rival could one day apply to AMD. The comeback is genuine. Whether it is durable is the question the next decade will answer.

Why This Matters to Investors

AMD is the answer to Intel, the proof that no moat is permanent and no giant is safe. Where Intel shows how a dominant company can lose the future by defending its comfort, AMD shows how a near-dead challenger can seize that future by concentrating everything on a single asymmetric bet. The Greatest Companies thesis is that durable competitive advantage builds lasting wealth, and AMD complicates it in a useful way. AMD’s rise was not built on a deep, unassailable moat. It was built on superb execution, a smart architecture, and a rival’s mistakes. That made the comeback spectacular and the wealth creation enormous, with the stock rising more than twenty-fold from the lows. But it also means AMD must keep running, because the very playbook that let it win is available to whoever comes next. The investor’s lesson is to distinguish between a great run, which AMD has had, and a durable moat, which is a separate and harder thing, and never to confuse a brilliant comeback with permanent safety. None of this is investment advice; it is a way of reading history.

Lessons in Order of Depth

Method: smart architecture beats a big budget

AMD did not beat Intel with money. It beat Intel with a better idea, building processors from smaller chiplets instead of giant monolithic dies, and by renting the world’s best manufacturing rather than owning second-best. The method is that an asymmetric strategy, attacking where the giant is structurally weak rather than matching it head-on, can let a smaller player win a fight it has no business winning on paper.

Money: survive first, then you can win

Before AMD could make its brilliant bet, it had to be alive to make it. Going fabless, taking the low-margin console business, cutting every distraction, all of it was about buying time, staying solvent long enough for the big bet to land. The trader’s parallel is exact. Survival is not a consolation prize, it is the precondition for everything. The investor who blows up cannot be there for the recovery, no matter how right the eventual thesis turns out to be.

Mind: the discipline is in the boredom

AMD’s comeback looks dramatic in summary, but lived day to day it was unglamorous and repetitive: a clear roadmap, executed on schedule, year after year, while resisting the temptation to chase every shiny distraction. Lisa Su’s gift was not a single flash of genius but a decade of consistency. The lesson is that the most powerful results usually come not from brilliance in a moment, but from discipline sustained across a very long time.

The deepest question: is a comeback a moat?

AMD proved that an underdog can topple a giant. But the deepest question its story raises is whether winning the last war buys any protection in the next one. AMD rose by being a nimble, fabless designer that out-executed a slow incumbent. Nothing stops a future challenger from doing exactly that to AMD. A comeback is an achievement, not a fortress. The hardest task in business is not climbing to the top, which AMD has done twice, but building something at the top durable enough that the next hungry challenger cannot simply repeat your own trick against you.

The Legendary Scorecard

Founder Vision	7
Innovation	9
Execution	9
Moat	6
Capital Allocation	8
Wealth Creation	9
Durability	7
Historical Importance	7
Overall	8.0

The overall figure is an editorial verdict, not a weighted average. AMD earns top marks for innovation, execution, and wealth creation, because the Lisa Su turnaround is one of the great feats of corporate management and made enormous sums for those who believed early. It scores lower on moat and durability, because its advantage rests on continued execution against fierce competition rather than on an unassailable structural position, and a comeback, however brilliant, is not the same as a fortress.

At a Glance

Timeline

• 1969: Jerry Sanders and seven Fairchild colleagues found AMD
• 1975: AMD enters the microprocessor market as an Intel rival
• 2003: AMD pioneers 64-bit x86 with Opteron and Athlon 64, forcing Intel to follow
• 2006: AMD buys the graphics company ATI for about $5.4 billion
• 2009: AMD spins off its factories into GlobalFoundries and goes fabless
• 2013: AMD chips power the PlayStation 4 and Xbox One, helping it survive
• 2014: Lisa Su becomes CEO with the company near bankruptcy
• 2017: the Zen-based Ryzen and EPYC processors launch, built at TSMC
• 2022: AMD surpasses Intel in market value and completes the roughly $49 billion Xilinx acquisition
• 2025: AMD’s data center revenue overtakes Intel’s data center group, and its Instinct accelerators challenge Nvidia in AI

Key Numbers

Related Reading

AMD’s story only makes sense against the company it spent its life chasing. Read how Intel built the deepest moat in technology and then watched it become a trap, the giant AMD finally overtook. See how Nvidia built the chips for the artificial-intelligence age that both AMD and Intel are now racing to catch, the new giant in AMD’s sights. And study how TSMC, the Taiwanese foundry, became the manufacturing partner whose factories made AMD’s entire comeback physically possible. For the full collection, see our Greatest Companies of All Time hub.

Go Deeper

AMD is a study in survival and asymmetric strategy, in how a near-dead underdog can out-think a giant by refusing to fight on the giant’s terms, and in why staying alive long enough for your bet to land matters more than the bet itself. Those are the same forces that separate traders who endure a drawdown and compound from those who blow up before the thesis pays, and the book teaches you the discipline to protect your survival first and concentrate your edge where it actually counts.

This article is part of the Greatest Companies series, adapted from the book Greatest Companies, now available on Kindle.

Frequently Asked Questions

When was AMD founded, and by whom?

Advanced Micro Devices was founded on May 1, 1969, in Sunnyvale, California, by Jerry Sanders and seven colleagues from Fairchild Semiconductor. It began making logic and memory chips and entered the microprocessor market in the mid-1970s, beginning a lifelong rivalry with Intel, which had been founded a year earlier by other Fairchild veterans.

How close did AMD come to bankruptcy?

Very close. By the mid-2010s AMD’s stock had fallen toward two dollars a share, it carried about two billion dollars in debt, its Bulldozer chip architecture had flopped, and analysts widely modeled the company as a likely bankruptcy. Semi-custom chips for game consoles and a fabless cost structure kept it alive long enough to recover.

Who turned AMD around?

Lisa Su, an MIT-trained engineer who became chief executive in late 2014. She refocused the company on high performance, killed distractions, and bet its survival on a clean-sheet processor architecture called Zen and a chiplet design approach, manufactured by TSMC. The Ryzen and EPYC processors that resulted, launched in 2017, drove the turnaround.

How did AMD overtake Intel?

AMD out-designed Intel with its chiplet architecture and rode TSMC’s manufacturing leadership while Intel stumbled on its own factories and missed key transitions. AMD steadily took share in PCs and the data center, and in 2022 its market value surpassed Intel’s for the first time. By 2025, in one quarter, AMD’s data center revenue overtook Intel’s data center group.

Has AMD beaten Nvidia in AI?

No. AMD’s Instinct accelerators have become the leading credible alternative to Nvidia in artificial intelligence, and major cloud companies want a second source, but Nvidia remains dominant. AMD is the challenger in AI, much as it once was the challenger to Intel, and that race is far from settled.