AMD Business Model

From memory chips to Intel's shadow

Jerry Sanders founded Advanced Micro Devices on May 1, 1969, in Sunnyvale, California, after leaving Fairchild Semiconductor with seven co-founders. The company built its early business selling memory chips, then pivoted toward microprocessors after signing a licensing agreement with Intel in 1975 that let it produce Intel-compatible chips. That agreement gave AMD a foothold in the personal computer market, though its first original microprocessor, the Am9080, failed commercially in 1976.

AMD spent much of the following two decades locked in a familiar position: technically capable, priced lower than Intel, but rarely ahead of it. The 2003 launch of the Opteron processor, the first 64-bit x86 chip on the market, briefly put AMD ahead on a specific technical front and helped it gain ground in servers and workstations.

1

The ATI bet and the lean years

AMD acquired Canadian graphics card maker ATI Technologies for $5.4 billion in 2006, a deal meant to combine processors and graphics into single platforms that could compete with Intel and Nvidia at once. The acquisition strained AMD's balance sheet for years afterward, and the company's stock price and market share both declined through the early 2010s as Intel's manufacturing lead widened.

In 2009, AMD spun off its chip manufacturing operations into what became GlobalFoundries, a move that let the company focus capital on chip design rather than the costly business of building and running fabrication plants. That decision, unpopular with some engineers at the time, positioned AMD to outsource future manufacturing to Taiwan Semiconductor Manufacturing Company as that foundry's process technology caught up to and eventually surpassed Intel's.

AMD spun off its chip manufacturing operations in 2009 to focus capital on design rather than factories.

2

Lisa Su and the Zen turnaround

Lisa Su became chief executive in October 2014, after AMD had cut headcount repeatedly and watched its stock fall below $2 a share. Su redirected engineering resources toward a new chip architecture called Zen, betting that a clean-sheet redesign could close the performance gap with Intel rather than incrementally improving AMD's aging Bulldozer architecture. The first Zen-based Ryzen processors launched in 2017, followed by EPYC server chips aimed directly at Intel's most profitable product line.

The bet paid off gradually rather than immediately. AMD's data center revenue grew slowly through the late 2010s before accelerating once EPYC processors demonstrated competitive performance and lower total cost of ownership for cloud providers such as Amazon Web Services and Microsoft Azure.

3

Gaming consoles as a steady licensing business

AMD supplies the custom processors inside Sony's PlayStation and Microsoft's Xbox consoles, a business built on semi-custom chip design rather than off-the-shelf parts. For the PlayStation 5, AMD engineers worked directly with Sony to combine a Zen 2 processor core with a custom RDNA graphics architecture, tailored to the console's specific performance and power targets. This work generates royalty and licensing revenue that is less cyclical than PC or graphics card sales, since console generations run for years once launched.

The relationship also gives AMD design wins it can reuse elsewhere, since techniques developed for console chips have fed into later PC and data center products.

Betting on the AI data center cycle

AMD's Instinct line of accelerator chips competes directly with Nvidia's GPUs for a share of the artificial intelligence training and inference market, though AMD remains the smaller player in that specific segment. The company signed a supply agreement with OpenAI in 2025 that positioned AMD as a key chip provider for the AI developer's infrastructure needs, a deal that contributed to record data center revenue in the back half of the year. AMD's fourth-quarter 2025 results showed data center revenue climbing sharply as cloud providers diversified their chip suppliers rather than depending solely on Nvidia.

Whether AMD can sustain that growth depends on continuing to ship competitive accelerator chips on a yearly cadence, matching the pace Nvidia has set with its own product roadmap. AMD's advantage lies less in raw performance than in offering large AI infrastructure buyers a credible second source, reducing their dependence on a single supplier for the chips that power AI models.

4

Ownership and leadership

As a publicly traded company on the Nasdaq exchange, AMD is owned by institutional and individual shareholders, with The Vanguard Group, BlackRock and State Street Global Advisors among its largest institutional holders. Lisa Su continues to serve as chair and chief executive, a role she has held since 2014 and one credited with steering the company from near-irrelevance to a central position in AI infrastructure supply chains.

5

Key Partners

Taiwan Semiconductor Manufacturing Company serves as AMD's primary foundry partner, producing chips designed in-house by AMD engineers. Original equipment manufacturers, including Dell, HP and Lenovo, integrate AMD processors and graphics cards into their own computers and servers. Cloud service providers such as Microsoft Azure and Amazon Web Services deploy AMD's EPYC and Instinct chips in their data centers, while Sony and Microsoft partner with AMD for custom console processors.

Key Activities

Chip design and architecture development form AMD's central activity, spanning processor cores, graphics architectures and the software that supports them. The company manages a complex supply chain that coordinates with foundry partners on manufacturing capacity and yield. Sales and marketing efforts target both consumer channels and large enterprise and cloud customers, while ongoing software development supports drivers and optimization tools that help customers extract full performance from AMD hardware.

Key Resources

AMD's intellectual property, including its Zen processor architecture and RDNA graphics architecture, underpins its entire product lineup and competitive position. A large engineering workforce concentrated in chip design and semiconductor physics represents a resource competitors cannot easily replicate. The company's relationships with foundry partners, particularly its long-term capacity agreements with Taiwan Semiconductor Manufacturing Company, secure the manufacturing access AMD needs without owning factories itself.

Value Propositions

For PC enthusiasts and gamers, AMD offers processors and graphics cards priced below comparable Intel and Nvidia products while delivering competitive performance. For data center customers, AMD's EPYC processors and Instinct accelerators promise lower total cost of ownership through energy efficiency and pricing relative to entrenched competitors. For console makers such as Sony and Microsoft, AMD provides custom semi-custom chip design that blends processor and graphics capability into a single package tailored to each console's requirements.

Customer Relationships

AMD maintains large enterprise and cloud customers through direct account management and technical engineering support, given the complexity of deploying processors and accelerators at data center scale. Consumer customers interact with AMD through its website, community forums and social media channels, where the company gathers feedback on product performance and software issues. Retail and OEM partners receive marketing co-investment and technical support to help sell AMD-powered products.

Channels

AMD sells directly to large OEMs, cloud providers and console makers through dedicated sales teams handling high-volume contracts. Retail partners and online marketplaces distribute AMD's consumer processors and graphics cards to individual buyers. The company's website and support forums serve as both an informational channel and a customer service touchpoint for troubleshooting and product guidance.

Customer Segments

PC enthusiasts and gamers make up a core consumer segment, purchasing processors and graphics cards for desktops and laptops. Data centers and cloud service providers represent AMD's fastest-growing segment, buying EPYC processors and Instinct accelerators for enterprise computing and AI workloads. OEMs and system integrators, along with enterprise, government and education customers, round out AMD's base, each valuing a different mix of performance, affordability and reliability.

Cost Structure

Research and development spending represents one of AMD's largest cost categories, reflecting the expense of designing new processor and graphics architectures on a multiyear cadence. Manufacturing costs, paid to foundry partners such as Taiwan Semiconductor Manufacturing Company, scale with production volume and the complexity of each chip's process node. Sales, marketing and administrative expenses support AMD's global operations, while acquisition-related costs recur periodically as the company buys smaller technology firms to fill capability gaps.

Revenue Streams

Processor and graphics card sales generate the bulk of AMD's revenue, sold to PC manufacturers, retailers and individual consumers. Data center products, spanning EPYC server chips and Instinct AI accelerators, form the company's fastest-growing revenue stream. Licensing and royalty income from semi-custom console chips, sold to Sony and Microsoft, provides a steadier revenue base less exposed to PC market cycles, alongside smaller intellectual property licensing income from other technology companies.

• 1AMD company history and founding
• 2AMD spins off manufacturing into GlobalFoundries
• 3Lisa Su profile and AMD turnaround
• 4AMD full year 2025 revenue reaches $34.6 billion
• 5AMD institutional ownership structure