For most of the 2000s, Advanced Micro Devices looked like a company that couldn’t get out of its own way. Intel owned the processor market with something approaching monopoly-level dominance, spending billions on fabrication plants, developer relationships, and the kind of marketing budget that put “Intel Inside” stickers on every laptop sold in the Western world. AMD was the scrappy alternative, perpetually underfunded, occasionally brilliant, and chronically unprofitable.
Then something strange happened. AMD stopped trying to beat Intel at Intel’s game, and started printing money.
The setup requires some history. AMD and Intel have shared a peculiar relationship since the 1980s, when IBM required Intel to license its x86 architecture to a second source as a condition of the original PC contract. AMD became that second source, which granted it permanent legitimacy in a market it didn’t create and couldn’t fully control. For decades, AMD competed directly on Intel’s terms: faster clock speeds, better benchmarks, lower prices. Sometimes AMD won (the Athlon 64 era, roughly 2003 to 2006, saw AMD genuinely embarrass Intel on performance). More often, Intel’s manufacturing advantages crushed AMD’s margins before the products even shipped.
The near-death experience came around 2012. AMD had lost its fabrication advantage after spinning off its fabs into what became GlobalFoundries. Its CPU designs were falling behind. Its GPU business, acquired through the ATI purchase in 2006, was hemorrhaging cash competing against Nvidia. The company’s stock hit lows that had analysts writing obituaries.
What happened next is the case study.
Lisa Su took the CEO role in 2014, and the strategic pivot she executed wasn’t primarily about making better chips (though AMD did make better chips). It was about redefining what winning meant. Intel’s dominance in consumer and enterprise CPUs was real and durable. Nvidia’s lead in discrete consumer GPUs was real and durable. AMD had neither the balance sheet nor the manufacturing capacity to dislodge either company from its core market.
So AMD stopped trying to.
Instead, AMD focused on three things that a market leader structurally cannot prioritize: serving customers the leader ignores, winning on price-to-performance rather than absolute performance, and moving faster in markets where agility matters more than scale.
The console market illustrates this perfectly. Sony and Microsoft both chose AMD processors and graphics for their current console generations (PlayStation 4, PlayStation 5, Xbox One, Xbox Series X/S). This was not accidental. Intel couldn’t offer the integrated CPU-GPU designs the consoles required at competitive prices, and Nvidia, burned by a difficult Sony relationship in the PS3 era, priced itself out. AMD, operating from a position of hunger rather than market power, offered custom silicon at margins that made the deals work. The result was a guaranteed, high-volume revenue stream that didn’t require AMD to beat anyone at anything.
The Ryzen launch in 2017 is where the financial logic became undeniable. AMD released processors that matched or exceeded Intel’s consumer CPUs at significantly lower price points. AMD didn’t need to be better than Intel across every segment. It needed to be good enough to capture the price-sensitive half of the market while Intel’s own pricing remained anchored to its premium positioning. A company with 30 percent market share and healthy margins can be more profitable than a company with 60 percent share that has spent years subsidizing that dominance.
This is the counterintuitive core of the second-place advantage: the market leader must defend everywhere. AMD only needed to win somewhere.
Intel, during this period, was trapped by its own success. Its tick-tock manufacturing cadence (alternating between new architectures and new process nodes) had stalled. The 10nm process node was years late. Intel’s response was to keep selling refined versions of aging architectures at premium prices because its enterprise customers, locked into Intel’s platform through software optimization and supply chain inertia, had nowhere else to go. This is the market leader’s curse: your installed base becomes your prison.
AMD had no installed base to protect. Every design decision was made without the weight of backward compatibility, existing customer commitments, or the need to not disrupt a profitable status quo. The Zen architecture, designed from scratch, could optimize for the performance-per-watt metrics that mattered to data center customers in 2017 rather than the clock-speed metrics that had mattered in 2005.
By 2020, AMD’s EPYC server processors were winning significant data center contracts at AWS, Microsoft Azure, and Google Cloud. Not because AMD had beaten Intel comprehensively, but because AMD had become good enough in exactly the dimensions hyperscalers care about (core count, memory bandwidth, power consumption) while undercutting Intel on price.
The GPU side of AMD’s business shows the same pattern against Nvidia. AMD’s Radeon cards have never consistently beaten Nvidia’s best, and AMD has never pretended they would. Instead, AMD has competed aggressively in the mid-range market where the volume is. It has also used its CPU-GPU integration expertise to win markets Nvidia can’t serve as cleanly: the console silicon, the APU market for thin-and-light laptops, and increasingly the data center AI accelerator space where AMD’s MI300 chips are positioned as a credible (if not dominant) alternative to Nvidia’s H100.
The financial results validate the strategy. AMD’s revenue grew from roughly $4 billion in 2016 to over $22 billion in 2022. Operating margins, once a source of embarrassment, reached levels competitive with Intel’s. The stock, trading under $5 in 2015, exceeded $150 at its 2021 peak.
This story connects to a broader principle that gets undervalued in technology analysis. Market leadership is expensive. The leader pays for category creation, for the sales force required to defend every segment, for the R&D required to maintain dominance on all fronts simultaneously. The second-place player gets to read the market map the leader drew and choose which territories are actually worth contesting.
As we’ve explored elsewhere on this site, the second-place dynamic appears across markets with surprising consistency. But AMD’s case is particularly instructive because the company spent so long trying to win on the leader’s terms before discovering that the more profitable game was playing on its own.
The lesson isn’t that you should try to stay second. It’s that if you are second, with less capital and less market power, the rational play is almost never a frontal assault. AMD spent fifteen years attempting that assault and nearly went bankrupt. It spent the next seven years picking its spots, and became a company worth over $150 billion.
Sometimes the most important strategic decision is choosing which fights you’re not going to have.
