Graphics processing units (GPUs) are ubiquitous and found in many of the devices that we interact with on a daily basis. In addition, GPUs are used for AI and supercomputer workloads, and high-end processors from companies like AMD or Nvidia can cost $10,000–$30,000 or more. Demand for these units is increasing rapidly. As a result, the revenue of GPU developers is growing, and this year the global GPU market is expected to exceed $98.5 billion, according to Jon Peddie Research.
Currently, there are 20 companies and seven IP vendors developing discrete, integrated, and embedded GPUs, JPR reports. Most of these graphics processors are entry-level integrated GPUs, and only a few companies develop standalone GPUs for gamers including the best graphics cards. However, most of the revenue is earned from GPUs that are not used for graphics: sales of AI and HPC GPUs total only several million units per year, but since they are sold for tens of thousands of dollars per unit, they bring in tens of billions for Nvidia and billions for AMD.
In just two quarters of its fiscal 2025, Nvidia earned $42 billion from its AI and HPC GPUs, and for the whole year, sales of its compute GPUs could exceed $90 billion. AMD expects sales of its AI GPUs to exceed $3 billion. It is hard to estimate the revenues of companies like Biren or MetaX, but we doubt their compute GPU sales come close to those of AMD.
