AMD

Advanced Micro Devices is the leading challenger to NVIDIA in AI accelerators, with its Instinct MI300, MI325, and MI350-series GPUs and the open ROCm software stack. It is also a dominant force in data-center CPUs through its EPYC line, which runs alongside GPUs in most AI servers. A landmark multi-gigawatt supply agreement with OpenAI in 2025 validated AMD as a credible second source for frontier-scale compute.

Founded in 1969, AMD spent decades as the underdog to Intel in CPUs before a turnaround under CEO Lisa Su, who took over in 2014, made its EPYC server chips and Ryzen processors technology leaders. That execution credibility carried into AI accelerators.

The Instinct MI300X and its successors compete with NVIDIA on memory capacity and inference price-performance, and AMD has invested heavily in the ROCm open software stack to lower switching costs from CUDA. Its 2022 acquisition of Xilinx added adaptive computing, and Pensando added data-center networking.

The defining 2025 event was a strategic deal to supply OpenAI with up to six gigawatts of GPUs, paired with a warrant giving OpenAI the option to take a large equity stake — a structure that tied a major buyer directly to AMD's success. Alongside supply deals with Oracle, Microsoft, and Meta, it established AMD as the clearest alternative to NVIDIA at scale.