AMD’s CEO, Lisa Su, has shared a bold vision for the future of the AI data center accelerator market, projecting it will grow to a massive $500 billion by 2028.
This surge in demand is largely driven by the increasing need for AI accelerators, particularly for inferencing applications, which are vital for various industries.
Su explained at the Advancing AI keynote that the data center accelerator market is currently experiencing an impressive 60% compound annual growth rate (CAGR), and she believes this momentum will continue in the coming years. With AI’s expanding role in areas like cloud applications, edge AI, and client AI, the demand for high-performance compute solutions is set to grow. AMD is positioning itself at the forefront of this revolution with several strategic initiatives aimed at strengthening its presence in the AI sector.
To capitalize on this growing demand, AMD is focusing on three main strategies: leadership in compute engines, building an open ecosystem, and providing full-stack solutions. The company recently unveiled its latest Instinct MI350 AI lineup, featuring the cutting-edge CDNA 3 architecture built on TSMC’s 3nm process node. These cards come equipped with a powerful HBM3E memory stack and are designed for high-performance workloads, with the flagship MI355X model drawing attention for its 1400W TDP
. AMD claims that its new offerings have reached parity with NVIDIA’s Blackwell GPUs in terms of performance, further boosting its competitive edge.
In addition to hardware advancements, AMD has also rolled out a new software ecosystem to support AI workloads. The ROCm 7 software stack introduces new algorithms and models, offering advanced features for scaling AI, as well as improved enterprise capabilities for large-scale deployments. These moves are part of AMD’s aggressive strategy to challenge NVIDIA, which has traditionally dominated the AI accelerator market.
Despite the competition, Su’s optimism about the future of AI in data centers is clear. With AI workloads becoming more diverse and computationally intensive, the market for accelerators is only set to expand. AMD’s investment in both hardware and software innovations positions the company as a key player in this rapidly growing field, which promises vast opportunities not just for AMD, but for other companies in the sector, including NVIDIA.