AMD Cannot Catch Nvidia, But It Sure Can Make Some Money
nextplatform.com Nov 4, 2026

AMD Cannot Catch Nvidia, But It Sure Can Make Some Money

AI-summarised brief · reviewed before publication

AMD’s datacenter GPU business remains far behind Nvidia’s, largely because AMD cannot secure enough high‑bandwidth memory (HBM) from Samsung, SK Hynix and Micron, limiting production of its Instinct MI‑series accelerators. While AMD’s Instinct MI‑400 chips load more HBM than Nvidia’s Blackwell and Rubin GPUs, the memory shortage forces AMD to sell fewer units and hampers its ability to take market share from Nvidia. AMD does, however, erode Intel’s position in server CPUs, now holding just over a third of X86 shipments and generating slightly less than half of its revenue in the first half of 2026. Revenue gaps have widened dramatically: Nvidia posted $26.9 billion in 2021 versus AMD’s $16.4 billion, $60.9 billion versus $22.7 billion in 2023, and forecasts project Nvidia at roughly $354 billion this year compared with AMD’s $52.2 billion, a gap that is expected to persist into 2029.

💡 Why It Matters

  • · AMD’s inability to scale HBM supply caps its AI‑accelerator growth, leaving Nvidia to dominate the lucrative generative‑AI market and shaping the competitive landscape for future data‑center investments.