AMD launches data center graphics card to challenge NVIDIA's AI chip dominance

AMD released a new graphics processing unit (GPU — the specialized chip that trains and runs AI models) designed for data centers, where companies like Google and Meta build large AI systems. This card directly competes with NVIDIA's market-leading offerings in a segment where NVIDIA controls roughly 80% of sales.

AMD's card is built to handle the same workloads as NVIDIA's — training massive language models (AI systems that understand and generate text) and running inference (the step where an AI produces answers to user queries) at scale. The hardware targets the same cloud providers and AI labs that currently depend on NVIDIA.

For businesses buying AI infrastructure, this matters because AMD's entry could create a second reliable supplier option, potentially lowering prices or improving delivery times. AI startups and enterprises building custom models may gain leverage in negotiations with chip makers for the first time in years.