Advanced Micro Devices, Inc., better known as AMD, is an American semiconductor company headquartered in Santa Clara, California. Its components power personal computers as well as servers, gaming consoles and industrial equipment. The company operates under a so-called “fabless” model: it designs its chips but outsources their manufacturing to specialized partners, notably TSMC. Its corporate and commercial website is amd.com.
A history of diversification and refocusing
Founded in 1969 by Jerry Sanders and several associates, AMD gradually established itself as an alternative to Intel in processors compatible with the x86 architecture. Its history has been marked by cycles of technological competition, financial difficulties and changes in business scope. The acquisition of graphics specialist ATI in 2006 brought it crucial expertise in GPUs, used for graphics, video games and, now, parallel computing.
In 2009, the spin-off of its manufacturing operations created GlobalFoundries and set in motion its transformation toward a design-focused model. Under the leadership of Lisa Su, who became CEO in 2014, AMD revamped its product lineup around the Zen processor architecture. The acquisition of Xilinx, completed in 2022, then expanded its portfolio into programmable circuits and adaptive computing, with applications in telecommunications, industry and aerospace.
From PCs to data centers
AMD’s operations revolve around several complementary product families. Ryzen processors target desktop and laptop computers, while EPYC processors address the needs of servers and cloud infrastructure. Radeon graphics cards and chips cater in particular to gaming and digital content creation. AMD also develops semi-custom chips, used in several generations of Microsoft and Sony consoles.
In data centers, Instinct accelerators target high-performance computing as well as the training and inference of artificial intelligence models. Their adoption depends as much on hardware performance as on the software environment. AMD is therefore developing ROCm, an open software platform designed to harness its GPUs for these uses. Technologies originating from Xilinx complement this offering with reconfigurable components suited to specialized processing and latency or power consumption constraints.
What’s next?
Growing investment in artificial intelligence is opening up opportunities for AMD, but the company must make a compelling case beyond its chips’ specifications. System availability, the maturity of software tools and developer support weigh heavily in customer decisions. When competing with Nvidia’s ecosystem, the ability to simplify deployment is a central challenge.
AMD must also contend with its dependence on its partners’ manufacturing and packaging capacity, as well as US export restrictions on certain advanced components. Energy efficiency, integration between processors and accelerators, and diversification of end markets will be decisive in translating demand into lasting market positions.